REMESYS TECH LEARNING SOLUTION · COMPLETE CURRICULUM
MASTER
ARTIFICIAL
INTELLIGENCE
ARTIFICIAL
INTELLIGENCE
India's most comprehensive AI education program. MIT & Harvard level curriculum covering everything from Python fundamentals to cutting-edge Agentic AI, Deep Learning, IoT, Robotics, and real-world domain applications across Healthcare, Defence, Finance, Agriculture, and more. NSQF Level 12–14 certified, Government of India approved, with a 3-month paid internship at Krenx Technologies Pvt. Ltd.
12
Months Max
800+
Lesson Hours
50+
Real Projects
70+
Tools Mastered
15+
AI Domains
3mo
Krenx Internship
CERTIFICATION
NSQF Level 12 to 14
All courses are aligned with the National Skills Qualifications Framework (NSQF) at advanced levels 12–14 — exceeding standard certifications, aligned with NiELIT, Government of India Ministry of Education, and international AI standards (IEEE, ISO/IEC 42001).
Basic · 3 Months
NSQF Level 12 · Foundation Python, Data Science, ML basics. Build 5 real projects.
NSQF 12
Intermediate · 6 Months
NSQF Level 12–13 · Deep Learning, Computer Vision, NLP, RAG, Agentic AI basics.
NSQF 12–13
Advanced · 9 Months
NSQF Level 13 · LLMs, Fine-tuning, Multi-agent, IoT+AI, Robotics, AI in domains.
NSQF 13
Adv. Diploma · 12 Months
NSQF Level 13–14 · Research-grade. AI warfare, AGI concepts, full MLOps, Krenx internship.
NSQF 13–14
🇮🇳 Government of India Approved. All Remesys courses are aligned with NSQF (National Skills Qualifications Framework), NiELIT standards, and the National Education Policy 2020 (NEP). Certificates are recognized for government jobs, higher education credit transfer, and internationally under mutual recognition agreements.
INCLUDED WITH EVERY COURSE
The Remesys Package
Exclusive T-Shirt + ID Card
Official Remesys merchandise. Wear your AI identity. Valid ID for student discounts at conferences and tech events.
State-of-Art LMS Access
World-class Learning Management System with video lessons, interactive code environments, quizzes, and progress tracking.
24/7 AI Doubt Clearing
Our custom AI tutor — trained specifically on our curriculum — is available round the clock for instant doubt resolution.
Parent Progress Dashboard
Real-time visibility into attendance, assignments, quiz scores, and overall learning trajectory for parents.
Cloud GPU Practice IDE
Browser-based IDE with GPU access for training deep learning models. No expensive hardware required.
Learn Anywhere, Anytime
All sessions recorded and available on LMS. Missed a class? Watch the full recording within hours. Global access with internet.
Remesys AI Model Access
Exclusive access to Remesys-trained AI models for education, code assistance, and project support throughout the course.
Krenx Internship (Diploma)
3-month industry internship at Krenx Technologies Pvt. Ltd. (krenx.in) for diploma students. Real work, real projects.
Placement Assistance
Resume building, mock interviews, GitHub portfolio review, and active employer connections for Intermediate+ students.
MULTIPLE DEMO CLASS PLANS
Demo Class Templates
Complete 60-minute demo class plans for different audiences. Each demo is a standalone, deeply educational experience that showcases Remesys quality. Instructor: Snehargha Ghosh.
01
DEMO CLASS 01 · CLASS 10–12 STUDENTS · 60 MIN
What is AI? — The Complete Beginner Experience
Perfect for school students with zero background. Takes them from "what is AI?" to running their first Python code in 60 minutes. Uses real-world Indian examples — Flipkart, ISRO, Cricket analytics, Jio AI.
Minute-by-Minute Plan
- 0–5 min: Hook — "You used AI 50 times today without knowing." Demo: Face unlock, YouTube recommendation, WhatsApp autocorrect.
- 5–12 min: What is AI, ML, DL, Gen AI — Visual diagram on screen. No jargon. Cricket bat analogy for neural networks.
- 12–20 min: Live ChatGPT / Claude demo — Ask it to write a poem in Hindi, explain quantum physics like a 10-year-old, write code.
- 20–30 min: LIVE PYTHON — Run first Python code on screen. print("Hello India!"), simple calculator, list of IPL teams.
- 30–40 min: Real AI demo — TensorFlow Playground. Students shout numbers, watch neural net learn live.
- 40–50 min: AI in India — ISRO using AI, BharatGPT, Ola/Uber, Zepto, Paytm fraud detection. Career paths, salary data.
- 50–58 min: Remesys course walkthrough — Courses, NSQF, benefits, pricing, Krenx internship.
- 58–60 min: Q&A + Enroll on spot with 40% discount offer.
Learning Outcomes
- Can define AI, ML, Deep Learning, and Generative AI in their own words
- Has run Python code for the first time — instant confidence boost
- Understands the career opportunity and salary landscape in India
- Knows the difference between ChatGPT, Claude, Gemini, and Llama
- Understands what NSQF certification means for their future
- Knows exactly what they'll build in the Remesys course
Interactive Elements: Live Python coding, TensorFlow Playground demo, live ChatGPT query, audience poll (what AI do you use daily?), hands-on "train a model" mini-activity with Google Teachable Machine.
02
DEMO CLASS 02 · ENGINEERING / COLLEGE STUDENTS · 60 MIN
From Data to Intelligence — ML + Deep Learning Deep Dive
For CS/ECE/EEE college students. Goes deeper — live Jupyter notebook, scikit-learn model training, neural network from scratch, discussion on transformers and GPT architecture.
Minute-by-Minute Plan
- 0–5 min: Hook — Live: "AI writes code better than most developers." Show GitHub Copilot completing a function in real-time.
- 5–15 min: The ML pipeline — data → features → model → prediction. Bias-variance tradeoff explained visually.
- 15–28 min: LIVE JUPYTER — Load Titanic dataset, EDA with pandas, train Random Forest, check accuracy. Students predict survival.
- 28–38 min: Neural Networks — Perceptron to deep net. ReLU, backprop, gradient descent animated. PyTorch: build a 3-layer net live.
- 38–48 min: Transformers & LLMs — "Attention is All You Need" explained in 10 min. How GPT-4 generates text token by token.
- 48–56 min: Career paths — AI Engineer, Researcher, MLOps, AI Product Manager. Salary data. Top companies hiring in India.
- 56–60 min: Course showcase + Q&A + enroll.
Tools Used Live in This Demo:
Jupyter Lab · NumPy · Pandas · Scikit-Learn · Matplotlib · PyTorch · TensorFlow Playground · Google Colab · GitHub Copilot · HuggingFace Spaces
Jupyter Lab · NumPy · Pandas · Scikit-Learn · Matplotlib · PyTorch · TensorFlow Playground · Google Colab · GitHub Copilot · HuggingFace Spaces
Key Concepts Covered
- Supervised vs Unsupervised vs Reinforcement Learning
- Overfitting, underfitting, cross-validation
- CNN architecture for image recognition
- Transformer architecture and self-attention
- LLM tokenization and next-token prediction
- Vector embeddings and semantic search
03
DEMO CLASS 03 · PROFESSIONALS / CAREER CHANGERS · 60 MIN
AI for Your Career — Build Real Products, Not Just Learn Theory
For working professionals wanting to upskill. Focuses on practical AI applications, building deployable products, business value of AI, and the Remesys-to-Krenx career pipeline.
Minute-by-Minute Plan
- 0–8 min: ROI of AI skills — salary comparison, job market data, companies hiring. "Your current skills + AI = 3× your market value."
- 8–20 min: LIVE BUILD — Build a working AI chatbot using OpenAI API + Streamlit in 12 minutes. Deployed on HuggingFace Spaces. Students can use it.
- 20–32 min: RAG demo — Upload a PDF (e.g. company policy doc), ask questions. Show how any business can build this. LangChain + ChromaDB.
- 32–42 min: Agentic AI — Live AutoGPT/CrewAI demo. An AI agent that searches, reads, summarises, and produces a report autonomously.
- 42–50 min: AI in your domain — Marketing AI, Healthcare AI, Finance AI, Logistics AI. Show real tools and salary packages.
- 50–58 min: Remesys + Krenx internship pathway. Weekend batches for professionals. Flexible timing.
- 58–60 min: Q&A + enroll.
Live Products Built:
1. AI Chatbot with OpenAI API + Streamlit (deployed live)
2. PDF Q&A with RAG + ChromaDB
3. Autonomous research agent with CrewAI
1. AI Chatbot with OpenAI API + Streamlit (deployed live)
2. PDF Q&A with RAG + ChromaDB
3. Autonomous research agent with CrewAI
Audience Takeaways
- Built 3 working AI apps in one session
- Understands the full Generative AI tech stack
- Knows how RAG works and its business applications
- Can articulate AI ROI to their manager or client
- Has a clear learning path for the next 12 months
- Connected with Krenx internship opportunity
04
DEMO CLASS 04 · PARENTS / DECISION-MAKERS · 45 MIN
Why Your Child Must Learn AI — A Parent's Guide
Designed for parents attending with their child. Non-technical, focused on future-proofing, career security, and the Remesys learning environment. Builds trust and confidence.
- 0–10 min: The AI revolution — what jobs will exist in 2030. Why AI literacy is the new English literacy. India's AI roadmap.
- 10–20 min: Show (not tell) — live demo of what a Remesys student builds. Medical AI, stock predictor, chatbot. Let child try it.
- 20–30 min: The Remesys environment — LMS tour, parent dashboard, 24/7 AI support, recorded classes, home practice tools.
- 30–40 min: Career outcomes — NSQF certificates, Krenx internship, placement support, salary ranges. Success stories.
- 40–45 min: Pricing, discount, admission process. Q&A with both parent and child.
Parent Dashboard Features Shown:
Real-time attendance tracking · Assignment scores · Quiz results · Project submissions · Weekly progress reports · Direct instructor messaging · Class recording access
Real-time attendance tracking · Assignment scores · Quiz results · Project submissions · Weekly progress reports · Direct instructor messaging · Class recording access
COMPLETE LESSON PLAN · MIT / HARVARD LEVEL
Full Curriculum
48 weeks of meticulously crafted lessons. Every week has theory, lab exercises, tool workshops, and mini-projects. Click any week to expand the full lesson plan. Abbreviations: T Theory L Lab/Hands-on P Project I IDE/Tool Q Quiz/Assessment D Demo/Guest Lecture R Research/Paper Reading
WEEK 01
Python Mastery — The Language of AI
From zero to confident Python programmer. Setup, syntax, data structures, functions, OOP.
20 hrs
▼
Day 1–2 · Environment & Core Syntax
THEORY
What is Python? Why AI engineers use Python over C++/Java. Ecosystem overview.Covers: interpreted vs compiled, dynamic typing, the Python philosophy
IDE
Install Python 3.12, VS Code, Jupyter Lab, Anaconda. Configure conda environments.Tools: Python, VS Code, Jupyter Lab, Anaconda, conda, pip
LAB
Variables, data types (int, float, str, bool, None), type casting, f-strings, print formatting.Hands-on: Build a student grade calculator
LAB
Operators (arithmetic, comparison, logical, bitwise), operator precedence, input().Project mini: Temperature converter Celsius ↔ Fahrenheit ↔ Kelvin
Day 3–4 · Data Structures Deep Dive
LAB
Lists: creation, indexing, slicing, list comprehension, map/filter/reduce, nested lists.Build: IPL team stats manager using lists
LAB
Tuples, Sets, Dictionaries — when to use each, performance implications.Build: Inventory management system using dicts
LAB
Control flow: if/elif/else, for/while loops, break/continue/pass, loop-else.Build: Number guessing game with attempts counter
Day 5 · Functions, Modules & OOP
THEORY
Functions: def, arguments, *args, **kwargs, default params, return, recursion, lambda.Why functions matter in ML pipelines — reusable, testable code
LAB
OOP: classes, objects, __init__, self, inheritance, polymorphism, encapsulation, dunder methods.Build: NeuralLayer class skeleton — preview of what's ahead
IDE
Python modules: import, from-import, as alias, __name__ == "__main__". Standard library tour.Tools: os, sys, math, random, datetime, json, pathlib, collections
QUIZ
Week 1 Assessment: Python Fundamentals — 30 questions + 2 coding challenges.Topics: data types, loops, functions, OOP basics
WEEK 02
NumPy — The Engine of Numerical AI
n-dimensional arrays, broadcasting, vectorised operations, linear algebra for ML.
18 hrs
▼
THEORY
Why NumPy? Python lists vs NumPy arrays — speed benchmarks, memory layout, vectorisation.Demo: 1M element sum — Python list (0.7s) vs NumPy (0.002s) — 350× faster!
IDE
NumPy arrays: ndarray, shape, dtype, ndim, size, itemsize. Creating arrays: zeros, ones, eye, linspace, arange, random.Tool: NumPy 1.26+, Jupyter Lab
LAB
Indexing and slicing: 1D, 2D, 3D arrays. Boolean indexing. Fancy indexing. np.where.Exercise: Filter all employees with salary > 50K from an array of records
LAB
Broadcasting rules explained with visual diagrams. Element-wise operations on different shaped arrays.Build: Batch normalisation formula from scratch using broadcasting
THEORY
Linear algebra for ML: dot product, matrix multiplication, transpose, inverse, determinant, eigenvalues.Why this matters: weights in neural networks ARE matrices
LAB
np.linalg: solve, svd, norm, lstsq. Principal Component Analysis (PCA) from scratch.Project: Implement linear regression using only NumPy (no sklearn)
LAB
Random module: seed, uniform, normal, choice, shuffle. Monte Carlo simulation.Project: Monte Carlo π estimation — 10M random points
PROJECT
NumPy Image Processing: Load an image as array, apply grayscale, blur, edge detection — all in NumPy.No OpenCV — pure NumPy convolution to understand CNN foundations
WEEK 03
Pandas — Data Wrangling at Scale
DataFrames, Series, EDA, data cleaning, merging, groupby — the bread and butter of data science.
20 hrs
▼
IDE
Pandas architecture: DataFrame, Series, Index. Reading data: read_csv, read_excel, read_json, read_sql, read_parquet.Tools: Pandas 2.x, SQLite, Parquet files
LAB
Indexing: loc, iloc, at, iat. Boolean filtering. .query() method. Multi-level indexing.Dataset: Real Zomato restaurant dataset — filter by city, rating, cuisine
THEORY
Exploratory Data Analysis (EDA) methodology — the 5-step framework used at Google, Netflix, and Amazon data teams.Shape/info → Nulls → Distribution → Correlation → Outliers
LAB
Data cleaning: handle missing values (dropna, fillna, interpolate), fix data types, string cleaning, regex in Pandas.Dataset: Messy Indian census data — clean and standardise
LAB
groupby, agg, pivot_table, crosstab, resample for time series. apply() and lambda.Dataset: IPL match-by-match data — player performance analytics
LAB
Merging/joining DataFrames: merge (inner, outer, left, right), concat, join. Database-style operations.Build: Customer 360 view by joining orders + customers + products tables
IDE
API data fetching: requests library, JSON parsing, storing in DataFrame. OpenWeatherMap API, Alpha Vantage stock API.Live fetch: Real NSE/BSE stock data into Pandas DataFrame
PROJECT
🚀 Mini Project: COVID-19 India Analysis Dashboard — fetch live data via API, clean, EDA, insights report.Output: Automated HTML report with 10 key findings
QUIZ
Pandas Assessment: 25 questions + live dataset challenge in 45 minutes.
WEEK 04
Data Visualisation — Making Data Speak
Matplotlib, Seaborn, Plotly, interactive dashboards. The art of communicating with data.
16 hrs
▼
IDE
Matplotlib deep dive: Figure, Axes, subplots, GridSpec. Line, bar, scatter, histogram, box, pie, heatmap.Customisation: fonts, colors, annotations, twin axes, log scale
IDE
Seaborn: statistical plots — displot, violinplot, boxplot, pairplot, heatmap, FacetGrid, jointplot, lmplot.Dataset: World Happiness Index — correlate factors with happiness scores
IDE
Plotly & Plotly Express: interactive charts, choropleth maps, 3D scatter, animated bubble charts.Build: Interactive India state-wise education and literacy dashboard
LAB
Streamlit: build interactive data web apps in pure Python. st.slider, st.selectbox, st.dataframe, st.plotly_chart.Deploy: Streamlit stock analysis app on Streamlit Cloud (free)
PROJECT
🚀 Project: Real Estate Insights Dashboard — Bangalore housing dataset, 8 interactive visualisations, deployed live.Tools: Pandas, Plotly, Seaborn, Streamlit, Streamlit Cloud
WEEK 05
Statistics & Probability for AI
The mathematical foundations every AI practitioner must know. Distributions, hypothesis testing, Bayesian thinking.
18 hrs
▼
THEORY
Descriptive stats: mean, median, mode, variance, standard deviation, skewness, kurtosis, IQR, percentiles.Why each metric matters in ML: outlier detection, feature scaling decisions
THEORY
Probability foundations: events, sample space, conditional probability, independence, Bayes' theorem.Application: Spam filter — naïve Bayes derivation from scratch
THEORY
Probability distributions: Normal, Binomial, Poisson, Bernoulli, Exponential, Uniform. When to use each.Visualise all distributions in Python with SciPy + Matplotlib
LAB
Central Limit Theorem — simulate with NumPy. Law of Large Numbers. Z-score, t-score, p-value.Hands-on: Sample 1000 random experiments, prove CLT live
LAB
Hypothesis testing: null/alternative hypothesis, t-test, chi-square test, ANOVA. scipy.stats module.Real case: Did Kohli's average improve after IPL 2022? Statistical significance test.
THEORY
Correlation vs causation, Pearson/Spearman correlation, R² interpretation, information entropy, KL divergence.Why KL divergence matters: loss functions in VAEs and GANs
LAB
Linear Algebra review: matrix multiplication, eigendecomposition, SVD. Applications in PCA and recommender systems.Implement: SVD-based movie recommender from scratch with NumPy
QUIZ
Math for ML Assessment: 30 questions covering probability, statistics, linear algebra.Include: derivation problems, Python code problems, interpretation questions
WEEK 06
Complete Data Science Pipeline — End-to-End
Real-world data pipeline: fetch → clean → analyse → visualise → API → deploy. Full Kaggle workflow.
20 hrs
▼
LAB
Kaggle platform mastery: datasets, kernels, competitions, submission format, leaderboard strategy.Set up Kaggle API, download datasets programmatically
IDE
Google Colab: GPU/TPU runtime, mounting Google Drive, sharing notebooks, colab secrets for API keys.Alternative: Jupyter Lab with Docker, Paperspace Gradient (free tier)
LAB
Web scraping with Beautiful Soup + Selenium + Playwright. Scrapy for large-scale crawling.Scrape: Naukri.com AI job listings → analyse skills in demand → visualise trends
IDE
SQL for data scientists: PostgreSQL + psycopg2 + SQLAlchemy. SELECT, JOIN, GROUP BY, window functions, CTEs.Connect Python → SQL database → Pandas → analysis pipeline
PROJECT
🚀 CAPSTONE PROJECT 1: Indian Agriculture Analytics PlatformFetch crop yield data from data.gov.in API → clean → EDA → 15 visualisations → Streamlit dashboard → deploy. Full professional pipeline.
PROJECT
🚀 CAPSTONE PROJECT 2: Financial Market DashboardFetch real NSE/BSE data via yfinance API → technical indicators (RSI, MACD, Bollinger Bands) → interactive Plotly charts → Streamlit app → share link.
WEEK 07
Machine Learning Fundamentals & Scikit-Learn
The ML pipeline, scikit-learn API, train/test split, cross-validation, bias-variance tradeoff.
20 hrs
▼
THEORY
Supervised vs Unsupervised vs Reinforcement vs Self-Supervised Learning. The ML landscape in 2025.Map all techniques to real-world use cases: classification, regression, clustering, generation
THEORY
The ML pipeline: problem definition → data → features → algorithm selection → training → evaluation → deployment.Why most ML projects fail (bad data, wrong metrics, no monitoring)
IDE
Scikit-Learn API: fit, predict, transform. Estimators, transformers, pipelines. sklearn.pipeline.Pipeline.Understand the design pattern — every algorithm shares the same interface
LAB
Train/test split, stratified split, cross-validation (k-fold, stratified k-fold, LeaveOneOut). sklearn.model_selection.Live: See how random split can bias results — why stratification matters
THEORY
Bias-variance tradeoff — the fundamental tension in ML. Underfitting, overfitting, the sweet spot.Visual: Learning curves — how to diagnose your model's problem
LAB
Linear Regression: gradient descent from scratch, then sklearn.linear_model.LinearRegression. OLS, closed-form solution.Dataset: House price prediction with 79 features — Ames Housing
LAB
Evaluation metrics: MSE, MAE, RMSE, R², adjusted R², MAPE. How to pick the right metric for your problem.Common mistake: optimising MSE when you should optimise MAE (why?)
PROJECT
🚀 Project: Bangalore Real Estate Price Predictor — full pipeline with Ridge regression + deployed Streamlit app.Include: cross-validation, learning curves, feature importance plot
WEEK 08
Classification Algorithms — From Logistic to Ensemble
Logistic Regression, Decision Trees, Random Forest, SVM, KNN, Gradient Boosting, XGBoost, LightGBM.
22 hrs
▼
THEORY
Logistic Regression: sigmoid function, log-loss, decision boundary, multiclass (OvR, softmax). Math derivation.Implement from scratch with NumPy → then compare with sklearn
THEORY
Decision Trees: information gain, Gini impurity, entropy, pruning, depth control. Visualise with graphviz.Why trees overfit and how to fix it — the journey toward Random Forest
THEORY
Ensemble methods: Bagging (Random Forest), Boosting (AdaBoost, Gradient Boost, XGBoost, LightGBM, CatBoost).Why ensembles win Kaggle: wisdom of crowds, bias reduction
LAB
SVM: maximal margin classifier, kernel trick (RBF, polynomial), SVR. Hyperparameter C and gamma.Dataset: Breast cancer classification — compare SVM vs RF vs XGBoost
IDE
XGBoost + LightGBM: installation, DMatrix/Dataset API, GPU training, early stopping, feature importance.Tools: xgboost, lightgbm, catboost
LAB
Classification metrics deep dive: confusion matrix, precision, recall, F1, ROC-AUC, PR-AUC. When to use each.Medical example: 99% accuracy but useless — the precision-recall tradeoff in disease detection
IDE
SHAP (SHapley Additive exPlanations) for model interpretability. Feature importance. Partial dependence plots.Tools: shap, eli5, lime — explain ANY black-box model
PROJECT
🚀 Project: Credit Card Fraud Detection System — XGBoost + SMOTE for class imbalance + SHAP explanations + API.Real Kaggle dataset: 284,807 transactions, 0.17% fraud. Achieve 99.5%+ AUROC.
WEEK 09–10
Unsupervised Learning, Feature Engineering & Pipelines
K-Means, DBSCAN, PCA, t-SNE, UMAP, feature selection, sklearn pipelines, hyperparameter tuning.
24 hrs
▼
THEORY
Clustering: K-Means, K-Means++, Mini-batch K-Means. Elbow method, silhouette score. DBSCAN for noise-robust clustering.Application: Customer segmentation for e-commerce — build 5 distinct customer personas
THEORY
Dimensionality reduction: PCA (mathematical derivation), t-SNE (perplexity, learning rate), UMAP (topology-preserving).Visualise MNIST 784-dimension → 2D with t-SNE. See clusters form live.
LAB
Feature Engineering: polynomial features, binning, log transform, target encoding, cyclical encoding (time, day of week).Hands-on: Engineer 30+ features from raw timestamps and categorical variables
LAB
Feature selection: SelectKBest, RFE (recursive feature elimination), Lasso for feature selection, mutual information.Start with 100 features, select the best 20 — compare model performance
IDE
sklearn Pipeline: chain preprocessing + model. ColumnTransformer for mixed types. Make production-grade ML code.Tools: sklearn.pipeline, sklearn.compose, sklearn.impute, sklearn.preprocessing
IDE
Hyperparameter tuning: GridSearchCV, RandomizedSearchCV, Optuna (Bayesian optimisation), Hyperopt.Tune XGBoost: 15 hyperparameters, 200 trials with Optuna — automated optimisation
PROJECT
🚀 Project: E-Commerce Customer Segmentation + Recommendation EngineK-Means clustering → segment customers → collaborative filtering (SVD) recommendations → Streamlit dashboard
PROJECT
🚀 Project: Anomaly Detection in IoT Sensor DataDBSCAN + Isolation Forest + AutoEncoder (preview of deep learning) for industrial sensor anomaly detection
WEEK 11–12
NLP Fundamentals & Time Series Forecasting
Text classification, sentiment analysis, NLTK, spaCy, TF-IDF. Time series: ARIMA, Prophet, LSTM preview.
24 hrs
▼
THEORY
NLP pipeline: tokenisation, stemming, lemmatisation, stop word removal, POS tagging, NER, dependency parsing.Tools: NLTK, spaCy 3.x, TextBlob, Gensim
THEORY
Text representation: Bag of Words, TF-IDF, N-grams, Word2Vec, GloVe, FastText. From sparse to dense vectors.Visualise: Word2Vec "king - man + woman = queen" — the magic of word embeddings
LAB
Sentiment analysis: VADER (rule-based), TextBlob, ML-based (Logistic + TF-IDF), BERT-based (preview).Dataset: 50,000 IMDB reviews + 100,000 tweets. Compare approaches.
PROJECT
🚀 Project: Indian News Fake/Real Classifier — TF-IDF + SVM + SHAP + Streamlit web app.Scrape real news, build dataset, train model, deploy. Share link.
THEORY
Time series: stationarity, autocorrelation (ACF/PACF), seasonal decomposition, ARIMA, SARIMA, exponential smoothing.Use case: Predict Electricity demand for Indian states
IDE
Prophet by Meta: trend + seasonality + holiday effects. Auto-ARIMA. statsmodels.Forecast: 12-month Indian stock market prediction with confidence intervals
PROJECT
🚀 Project: Demand Forecasting for Retail Chain — predict inventory needs 30 days ahead for 50 products.Beats naive model by 40%. Business impact: ₹2Cr savings in overstocking.
WEEK 13–14
Model Deployment — FastAPI, Docker & Cloud
Take your model from Jupyter to production. REST APIs, Docker containers, cloud deployment, CI/CD basics.
22 hrs
▼
THEORY
ML deployment landscape: REST API, batch inference, streaming inference, edge deployment. Tradeoffs.Latency vs throughput: real-time (fraud detection) vs batch (monthly reports)
IDE
FastAPI: endpoints, Pydantic validation, async, background tasks, OpenAPI docs. Build ML model API in 30 min.Deploy: House price predictor as REST API with /predict endpoint
IDE
Docker: Dockerfile, images, containers, docker-compose, volumes, networking. Containerise ML models.Build: FastAPI + scikit-learn model in Docker container → test locally
IDE
Cloud deployment: HuggingFace Spaces (free), Render, Railway, Google Cloud Run, AWS Lambda for ML.Deploy 3 models to cloud — all free tier. Share live links.
LAB
Model versioning with pickle, joblib, ONNX. MLflow for experiment tracking and model registry.Tools: MLflow, pickle, joblib, ONNX Runtime
PROJECT
🚀 PHASE 2 CAPSTONE: Disease Prediction API Platform5 disease models (diabetes, heart, kidney, liver, stroke) → FastAPI → Docker → deployed on HuggingFace Spaces → React frontend. Full production system.
WEEK 15–16
Neural Networks from Scratch + TensorFlow & PyTorch
Perceptron → MLP → Deep Net. Backpropagation math. TensorFlow 2.x Keras API. PyTorch tensors and autograd.
28 hrs
▼
THEORY
Biological neuron → McCulloch-Pitts model → Perceptron (1957) → MLP → Deep Learning revolution (2012 AlexNet moment).History: Why deep learning failed in the 90s and why it succeeded in 2012
LAB
Build a 3-layer neural net from scratch in NumPy only: forward pass, loss (MSE, cross-entropy), backward pass, weight update.Train on XOR problem — the classic that stumped AI in the 70s
THEORY
Activation functions: Sigmoid, Tanh, ReLU, Leaky ReLU, ELU, GELU, Swish, Mish. When to use which.Plot all activation functions and their derivatives — understand vanishing gradient
THEORY
Optimisers: SGD, Momentum, Nesterov, AdaGrad, RMSprop, Adam, AdamW, LAMB. Math derivation.Visualise optimisers on a 3D loss landscape — see why Adam wins most of the time
THEORY
Regularisation: L1/L2 weight decay, Dropout (and variants), Batch Normalisation, Layer Normalisation, Early Stopping.Why BatchNorm was a breakthrough — smooth loss landscape, faster training
IDE
TensorFlow 2.x: tf.Tensor, eager execution, Keras Sequential → Functional → Subclassing API. Custom layers, losses, metrics.TensorBoard: visualise training curves, histograms, computation graphs
IDE
PyTorch: torch.Tensor, autograd, nn.Module, nn.functional, DataLoader, Dataset, Lightning.Build same MNIST classifier in both TF and PyTorch — compare APIs
LAB
Train on MNIST (99.2% acc), Fashion-MNIST, CIFAR-10. Learning rate schedulers: cosine annealing, warmup.Profile GPU usage. Understand batch size vs training speed tradeoff.
DEMO
TensorFlow Playground live session — students adjust architecture and hyperparameters in real time.Collaborative: whole class works together to get 0% training error on a spiral dataset
WEEK 17–18
Convolutional Neural Networks — Computer Vision
CNN architecture, famous models (VGG, ResNet, EfficientNet), transfer learning, object detection with YOLO.
28 hrs
▼
THEORY
How CNNs work: convolution operation, feature maps, filters, padding, stride, pooling (max, avg, global). Receptive field.Visualise: See exactly what each filter learns — edges, textures, shapes, objects
THEORY
CNN architectures: AlexNet (2012) → VGG (2014) → GoogLeNet → ResNet skip connections (2015) → DenseNet → EfficientNet → ConvNeXt.Read: "Deep Residual Learning for Image Recognition" — the most cited AI paper
LAB
Build ResNet-50 in PyTorch from scratch: residual blocks, batch norm, skip connections. Train on CIFAR-100.Reach 78%+ accuracy — beat vanilla CNN by 15%
LAB
Transfer Learning with TensorFlow Hub + PyTorch Hub: freeze base → train head → fine-tune all layers. EfficientNetV2.Build custom classifier: 10 Indian plant species recognition in 1 hour of training
IDE
OpenCV: image I/O, color spaces, geometric transforms, feature detection (SIFT, ORB), video capture, face detection (Haar + DNN).Tools: OpenCV 4.x, Pillow, Albumentations (data augmentation)
LAB
Object Detection: YOLO v8/v11 training — custom dataset annotation (Roboflow) → train → export → inference.Train on custom dataset: detect PPE (hard hat, vest) in construction site images
LAB
Image Segmentation: UNet for semantic segmentation. Instance segmentation with YOLOv8-seg.Segment: MRI brain tumour regions, road lanes for autonomous vehicles
PROJECT
🚀 Project: Medical Imaging AI — Chest X-Ray Pneumonia + COVID DetectionDenseNet121 + Grad-CAM visualisation → deployed API → web interface. Achieves 97.3% AUC.
PROJECT
🚀 Project: Real-Time Traffic Violation Detection SystemYOLOv8 → detect vehicles, helmets, license plates → alert system → live video stream
WEEK 19–20
Sequence Models — RNNs, LSTMs, GRUs, Attention
Sequential data mastery. Language models, time series with neural networks, the road to transformers.
24 hrs
▼
THEORY
RNN: unrolled through time, vanishing/exploding gradient problem, BPTT (backprop through time).Demo: RNN fails to remember beyond 10 steps — the memory crisis
THEORY
LSTM: cell state, forget gate, input gate, output gate. Mathematical derivation. The solution to vanishing gradient.GRU: simplified LSTM. When to use LSTM vs GRU vs Transformer
LAB
Build LSTM for stock price prediction (NIFTY 50) — 10 years of historical data → predict next 30 days.Compare: ARIMA vs Prophet vs LSTM vs Transformer
LAB
Sequence-to-Sequence: Encoder-Decoder LSTM for English → Hindi machine translation (small vocab).Understand: This is how early Google Translate worked (pre-transformer)
THEORY
Attention mechanism (Bahdanau, 2014): why attention fixed Seq2Seq. Query, Key, Value intuition.Read paper: "Neural Machine Translation by Jointly Learning to Align and Translate"
LAB
Temporal Convolutional Networks (TCN): dilated convolutions for sequences. Faster than LSTM.Apply: Anomaly detection in IoT time-series sensor data from a factory
PROJECT
🚀 Project: Hinglish Social Media Sentiment Analyser — LSTM with attention + real Twitter/X data.Handle code-switching (Hindi + English mixed). Deploy as API.
WEEK 21–22
Transformers — The Architecture Behind All Modern AI
"Attention is All You Need" in full depth. BERT, GPT architecture, positional encoding, multi-head attention, HuggingFace.
28 hrs
▼
THEORY
Transformer architecture: tokenisation → embeddings → positional encoding → multi-head self-attention → FFN → layer norm → output.Read: "Attention is All You Need" (Vaswani et al., 2017) — line by line
THEORY
Self-attention math: Q·Kᵀ/√d_k → softmax → ·V. Scaled dot-product. Multi-head parallelism. Causal masking for GPT.Implement: Self-attention from scratch in 50 lines of PyTorch
THEORY
BERT (encoder-only): masked language modelling, next sentence prediction, [CLS] token, fine-tuning paradigm.GPT (decoder-only): causal LM, next-token prediction, why this scales to ChatGPT
IDE
HuggingFace Transformers: AutoTokenizer, AutoModel, pipeline(), Trainer API, datasets library.Tools: transformers, datasets, tokenizers, evaluate, accelerate, peft
LAB
Fine-tune BERT for Indian legal document classification. Fine-tune mBERT for Hindi NER (named entity recognition).Beat baseline by 12% using pre-trained representations vs training from scratch
LAB
Vision Transformer (ViT) — apply Transformers to images. CLIP — connect images and text.Build: Image search engine using CLIP embeddings — search images with text queries
RESEARCH
Paper reading: "BERT" (Devlin 2018), "GPT-3" (Brown 2020), "InstructGPT" (Ouyang 2022), "LLaMA" (Touvron 2023).Understand: The trajectory from transformers → BERT → GPT → ChatGPT → Claude → Gemini
PROJECT
🚀 Project: Multilingual Customer Support System — BERT fine-tuned on Hindi + English + Tamil queries.Intent classification + entity extraction → route to right department + auto-suggest reply
WEEK 23–24
Large Language Models — How They Work & How to Use Them
LLM internals, tokenisation, context windows, GPT-4/Claude/Gemini APIs, local LLMs with Ollama.
26 hrs
▼
THEORY
LLM training pipeline: pre-training on web scale data → instruction tuning (SFT) → RLHF → Constitutional AI. Scaling laws.How Anthropic trains Claude differently from OpenAI's GPT — safety alignment
THEORY
Tokenisation: Byte Pair Encoding (BPE), WordPiece, SentencePiece. Context windows: 4K → 8K → 128K → 1M tokens.Why "needle in a haystack" problems emerge at long context
IDE
OpenAI Python SDK: Chat Completions, function calling, tools, streaming, vision, embeddings, batch API.Tools: openai 1.x SDK, Anthropic SDK, Google Generative AI SDK
IDE
Anthropic Claude API: claude-sonnet-4, claude-opus, system prompts, tool use, streaming, document analysis.Compare: Claude vs GPT-4o vs Gemini — strengths, pricing, rate limits
IDE
Ollama: run Llama 3.1, Mistral 7B, Phi-3, Gemma 2, CodeLlama locally. REST API + Python integration.Tools: Ollama, LM Studio, Jan, GPT4All — run AI on your laptop
LAB
Build: Multi-model comparison tool — same prompt → 5 different LLMs → compare speed, quality, cost.Benchmark on: creative writing, coding, reasoning, multilingual, math
PROJECT
🚀 Project: Multilingual AI Teacher AssistantAnswers class 10–12 questions in Hindi/English/Tamil via Claude API. Detects subject, adjusts difficulty, provides step-by-step solutions with diagrams (text-based). Deploy on WhatsApp Business API.
WEEK 25
Prompt Engineering — The New Programming
Zero-shot, few-shot, CoT, ReAct, Tree of Thought, meta-prompting, structured output, DSPy.
20 hrs
▼
THEORY
Zero-shot, one-shot, few-shot prompting. Instruction following vs in-context learning.A/B test: see how 3 examples in prompt improve accuracy by 25%
THEORY
Chain-of-Thought (CoT): "Let's think step by step." Why it works — forces the model to reason before answering.Self-consistency: sample 10 CoT paths → majority vote → more reliable answers
THEORY
ReAct: Reason + Act. Tree of Thought (ToT). Program-Aided Language Models (PAL). Least-to-most prompting.These techniques power AI agents — understanding them is critical
LAB
Structured output: JSON mode, function calling, Pydantic + instructor library. Parse LLM output reliably.Build: Resume parser that extracts structured JSON from any CV format
IDE
DSPy: programmable prompting. Optimise prompts automatically. BootstrapFewShot, BayesianSignatureOptimiser.Tool: DSPy — Stanford's framework for systematic prompt optimisation
PROJECT
🚀 Project: Automatic UPSC/JEE Answer Evaluator — CoT + structured output + scoring rubric.Evaluate 100 student answers automatically. Compare to human grader — 89% agreement.
WEEK 26–27
RAG — Retrieval Augmented Generation
Vector databases, embeddings, semantic search, hybrid search, advanced RAG patterns, production RAG.
28 hrs
▼
THEORY
Why RAG? LLM knowledge cutoff, hallucination, domain specificity. RAG vs fine-tuning — when to use which.RAG architecture: Indexing pipeline + retrieval + augmentation + generation
THEORY
Vector embeddings: dense representation of meaning. Cosine similarity, dot product similarity. Embedding models: text-embedding-3-large, BAAI/bge, E5, Nomic.Visualise: "cat" and "kitten" are close. "cat" and "car" are far.
IDE
Vector databases: ChromaDB (local), Pinecone (cloud), Weaviate, Qdrant, FAISS. CRUD operations, metadata filtering.Performance test: FAISS on 1M vectors — 10ms similarity search
LAB
Build Naive RAG: PDF loader → chunk → embed → store → retrieve → generate. LangChain + ChromaDB + Claude.Chat with your own textbook — upload Class 12 NCERT Physics, ask questions
LAB
Advanced RAG: HyDE, multi-query retrieval, contextual compression, re-ranking with CrossEncoder, parent-child chunking.Improve naive RAG answer quality by 35% with these techniques
LAB
Hybrid search: BM25 (keyword) + dense retrieval (semantic) combined with reciprocal rank fusion (RRF).Always better than either alone — the production standard
IDE
LlamaIndex (formerly GPT-Index): alternative to LangChain for RAG. Knowledge graphs, multi-modal RAG.Tools: llama-index, LangChain, Haystack, txtai
PROJECT
🚀 Project: Indian Legal Document AI AssistantIndex 5000+ IPC sections + SC judgements → RAG + Claude → answer legal queries in plain English/Hindi. Include confidence scoring + source citation.
PROJECT
🚀 Project: Hospital Knowledge Base ChatbotRAG over medical literature + hospital protocols → patient FAQ chatbot → deployed on WhatsApp via Twilio
WEEK 28
LLM Fine-Tuning — LoRA, QLoRA, Unsloth & RLHF
Customise open-source LLMs for your domain. Parameter-efficient fine-tuning on consumer hardware.
22 hrs
▼
THEORY
Fine-tuning landscape: Full fine-tuning vs PEFT (LoRA, QLoRA, Prefix Tuning, Prompt Tuning, IA³).Cost comparison: Fine-tuning GPT-4 ($1M) vs QLoRA on Llama 3 ($50 on Colab)
THEORY
LoRA: Low-Rank Adaptation. Only 0.1% of parameters trained. Math: W' = W + AB where rank(AB) << rank(W).QLoRA: quantised base model + LoRA adapters = fine-tune 70B model on 48GB GPU
IDE
Unsloth: 2× faster, 70% less memory. Fine-tune Llama 3.1, Mistral, Phi-3 on free Colab T4 GPU.Tools: unsloth, peft, trl, bitsandbytes, accelerate, transformers
LAB
Fine-tune Llama 3.1 8B on Indian legal data → domain expert LLM → compare before/after on legal Q&A benchmark.Dataset creation: Use GPT-4 to generate 5000 QA pairs from legal documents (synthetic data)
THEORY
RLHF: Supervised Fine-Tuning → Reward Model → PPO Optimisation. DPO (Direct Preference Optimisation) — simpler alternative.How Claude is made safe: Constitutional AI + RLAIF (RL from AI Feedback)
PROJECT
🚀 Project: BharatGPT Mini — Fine-tuned Llama 3 for Indian languages, culture, and domainsDataset: Indian recipes, Bollywood QA, cricket statistics, UPSC prep. Evaluate on custom benchmark.
WEEK 29–30
LangChain & LangGraph — Building AI Applications
Chains, agents, tools, memory, LangGraph stateful workflows. The production standard for LLM apps.
26 hrs
▼
THEORY
LangChain architecture: chains, agents, tools, memory, callbacks, LangSmith tracing.When to use LangChain vs direct API calls vs LlamaIndex
IDE
LangChain Expression Language (LCEL): composable pipelines with | operator. Runnables, RunnableParallel, RunnableBranch.Build: Parallel summarisation + fact-check + sentiment pipeline
LAB
LangChain agents: ReAct agent, tool calling. Build tools: web search (Tavily/DuckDuckGo), Python REPL, file I/O, custom APIs.Build: Research assistant that searches web + reads PDFs + writes structured report
IDE
LangGraph: directed graphs for stateful AI. State, nodes, edges, conditional routing, human-in-the-loop, persistence.Build: Customer support bot that can escalate, refund, or route to human agent
IDE
LangSmith: trace every LLM call, debug, evaluate, run CI for prompts. Annotate outputs, share traces.Essential for production: impossible to debug multi-step agents without tracing
PROJECT
🚀 Project: Autonomous Research & Report Writing AgentLangGraph agent: receives topic → plans research → searches 10 sources → reads full texts → synthesises → writes 2000-word report with citations → emails PDF report
WEEK 31–32
Multi-Agent Systems — CrewAI, AutoGen & Agent Orchestration
Teams of specialised AI agents working together. Role-based agents, hierarchical control, agent communication.
26 hrs
▼
THEORY
Multi-agent architectures: sequential, hierarchical, collaborative, competitive. When and why to use multiple agents.Analogy: AI team = CEO agent + CTO agent + Designer agent + QA agent
IDE
CrewAI: define agents with roles/goals/backstory, tasks, crew, process (sequential/hierarchical). Flows for complex orchestration.Build: Content creation crew — Researcher + Writer + Editor + SEO Specialist
IDE
Microsoft AutoGen: conversational multi-agent. AssistantAgent + UserProxyAgent + GroupChat. Code execution in Docker.Build: Coding team — AutoGen agents that write, test, debug, and document code
LAB
Agent memory: short-term (conversation), long-term (vector DB), entity memory. Agent-to-agent communication protocols.Build: Personal AI assistant that remembers your preferences across sessions
PROJECT
🚀 Project: AI Startup Simulator5 agents: CEO (strategy), CTO (tech decisions), CFO (budget), Marketing (campaigns), Sales (outreach). Give a product idea → agents build a full go-to-market plan + technical spec + financial model.
PROJECT
🚀 Project: Autonomous Software Engineering AgentInspired by Devin AI: give a GitHub issue → agent reads codebase → writes fix → runs tests → creates PR. Uses AutoGen + Claude + Docker sandbox.
WEEK 33–34
Image Generation & Multimodal AI
Stable Diffusion, DALL·E 3, ComfyUI, ControlNet, Sora (video), GPT-4o vision, building multimodal products.
24 hrs
▼
THEORY
Diffusion models: forward process (add noise) → reverse process (denoise). DDPM, DDIM, score matching. UNet backbone.How Stable Diffusion goes from random noise to a photorealistic image in 20 steps
THEORY
Latent Diffusion Models (LDM): compress to latent space first → diffuse there → decode. Why it's 8× faster than pixel-space diffusion.CLIP conditioning: how text prompts guide image generation (cross-attention)
IDE
Stable Diffusion: A1111, ComfyUI workflows, ControlNet (depth, canny, pose, seg), LoRA for style transfer, IP-Adapter.Tools: diffusers, ComfyUI, AUTOMATIC1111, Fooocus, InvokeAI
IDE
OpenAI DALL·E 3 + Ideogram + Flux via API. Image editing: inpainting, outpainting, style transfer.Build: Automated social media image generator from blog posts
LAB
Multimodal models: GPT-4o vision (describe images, OCR, document analysis), Claude 3.5 Sonnet (vision), Gemini 1.5 Pro (video).Build: Invoice/receipt extractor using vision LLM → structured JSON → database
LAB
Audio AI: Whisper (speech to text), ElevenLabs (text to speech), MusicGen, AudioCraft. Voice-first applications.Build: Voice-controlled home automation system prototype
PROJECT
🚀 Project: AI Creative Studio PlatformFull-stack app: text → image (Stable Diffusion) + text → audio (MusicGen) + text → video (Kling/Runway API) + style transfer + ControlNet. Deployed as SaaS.
WEEK 35
IoT Fundamentals + Edge AI
Sensors, actuators, MQTT, Raspberry Pi, Arduino, TensorFlow Lite, ONNX Runtime on edge devices.
24 hrs
▼
THEORY
IoT architecture: perception layer (sensors) → network layer (MQTT, CoAP, HTTP) → processing layer (edge/cloud) → application layer.Protocols: MQTT vs HTTP vs CoAP vs WebSocket for IoT
IDE
Raspberry Pi 5: GPIO, camera module, I2C, SPI, UART. Python GPIO library. Real sensor interfacing.Connect: temperature sensor, motion sensor (PIR), ultrasonic distance sensor, camera
IDE
Arduino + MicroPython on ESP32/ESP8266. WiFi+BLE. MQTT broker (Mosquitto). Node-RED visual IoT flows.Build: Smart home prototype — temperature + humidity → MQTT → Node-RED → dashboard
THEORY
Edge AI: why run AI on device? Latency, privacy, offline operation, bandwidth cost. TinyML overview.Edge vs cloud: real-time face unlock (edge) vs monthly analytics (cloud)
IDE
TensorFlow Lite: convert TF model → .tflite → quantise → deploy on Raspberry Pi/Android/microcontroller.ONNX Runtime: framework-agnostic inference on edge. OpenVINO for Intel hardware.
LAB
Deploy YOLOv8 on Raspberry Pi 5: real-time object detection at 15+ FPS. Optimise with FP16 + INT8 quantisation.Application: Smart doorbell that detects known faces vs strangers vs packages
PROJECT
🚀 Project: Smart Factory IoT + AI Quality Control SystemRaspberry Pi camera → TFLite defect detector → alert via MQTT → dashboard → auto-reject defective items. Industrial-grade prototype.
WEEK 36–37
Robotics with AI — ROS2, Path Planning & Computer Vision
Robot Operating System 2, autonomous navigation, AI-powered manipulation, simulation with Gazebo.
28 hrs
▼
THEORY
Robotics fundamentals: kinematics, dynamics, degrees of freedom, coordinate frames, transforms (TF2).Types: Industrial robots, collaborative robots (cobots), mobile robots, drones, humanoid robots
IDE
ROS2 (Robot Operating System 2): nodes, topics, services, actions, parameters, launch files. rclpy (Python API).Tools: ROS2 Humble, Gazebo, RViz2, Nav2, MoveIt2
LAB
Simulate a mobile robot in Gazebo: URDF model, differential drive, LIDAR, camera. Move it with velocity commands.Build: TurtleBot3 simulation — autonomous room navigation
THEORY
Path planning: A* algorithm, RRT, RRT*, Dijkstra, DWA (Dynamic Window Approach). Occupancy grid maps.SLAM: Simultaneous Localisation and Mapping. GMapping, Cartographer, ORB-SLAM3.
LAB
Nav2 stack: costmaps, planner server, controller server, recovery behaviours. Fully autonomous navigation.Robot navigates from room A to room B, avoiding dynamic obstacles
LAB
AI in robotics: integrate YOLOv8 with ROS2 — robot sees and identifies objects, moves toward target.Reinforcement Learning for robotics: OpenAI Gym + PyBullet + PPO policy
THEORY
Robotic arm control: inverse kinematics, MoveIt2 motion planning, collision avoidance, grasp planning with vision.Application: Pick-and-place robot guided by computer vision
PROJECT
🚀 Project: Autonomous Medical Delivery Robot (Simulation)ROS2 + Nav2 + YOLOv8 + SLAM. Navigates hospital corridors autonomously, identifies patient rooms, delivers medication. Full simulation in Gazebo + RViz2.
WEEK 38
AI in Warfare, Defence & DARPA-Level Systems
Autonomous weapons systems, AI in surveillance, swarm robotics, counter-drone systems, ethical dimensions.
16 hrs
▼
THEORY
Military AI landscape: autonomous weapons systems (AWS), lethal autonomous weapons (LAWS), UN debates, Geneva Convention applicability.Case studies: Israel's Harpy drone, US Navy Sea Hunter, Russia's Uran-9
THEORY
DARPA programs: OFFSET (swarm autonomy), AIR COMBAT EVOLUTION (AI dogfighting), Predictive Health and Readiness.AlphaDogfight Trials 2020: AI defeated human F-16 pilot 5–0. What this means.
THEORY
Swarm robotics: emergent behaviour, decentralised control, pheromone-based communication, swarm intelligence algorithms.DARPA OFFSET: 250 autonomous drones + ground robots operating as coordinated swarm
THEORY
AI in surveillance: CCTV + face recognition (SkyNet in China), predictive policing, gait recognition, crowd density analysis.Counter-AI: adversarial attacks on recognition systems. Deepfake detection for OSINT.
THEORY
Counter-drone systems: RF detection, radar, acoustic sensors, laser interception, AI-guided net capture. India's Raphael system.DRDO's AI initiatives: MPATGM, AURA UCAV, AI-based missile guidance
THEORY
Cybersecurity AI: AI-powered threat detection, autonomous penetration testing, deepfake-based social engineering, AI in electronic warfare.Ethical red lines: Geneva Convention, Martens Clause, autonomous kill decisions
GUEST
Guest Lecture: Defence technology expert / DRDO research perspective on India's AI in defence roadmap.Discussion: Should AI be allowed to make autonomous kill decisions? Class debate.
WEEK 39
AI in Healthcare & Precision Medicine
Medical imaging, drug discovery, genomics, clinical NLP, predictive health, digital twins for patients.
20 hrs
▼
THEORY
Medical imaging AI: radiology (X-ray, CT, MRI, PET), pathology (digital slides), ophthalmology (retinal scans), dermatology.FDA-approved AI devices: IDx-DR, Viz.ai, Caption AI — real deployed systems
LAB
Build chest X-ray AI: DenseNet-121 → Grad-CAM heatmap → clinical report generation with LLM → API → webapp.Dataset: NIH ChestX-ray14 (112,120 images, 14 pathologies)
THEORY
Drug discovery with AI: AlphaFold2 (protein structure), molecular property prediction, de novo drug design, clinical trial optimisation.AlphaFold solved 50-year protein folding problem in 2020 — the biggest AI breakthrough ever
LAB
Clinical NLP: extract diagnoses, medications, dosages, symptoms from unstructured clinical notes using BioBERT + spaCy.De-identification: remove PII from medical records automatically (HIPAA compliance)
PROJECT
🚀 Project: AI-Powered Rural Telemedicine SystemPatient describes symptoms in Hindi voice → Whisper ASR → LLM differential diagnosis → urgency triage → doctor video call scheduling. Built for India's 600,000 villages.
WEEK 40
AI in Education, Marketing, Finance & Logistics
Personalised learning, AI marketing, FinTech AI, supply chain optimisation — four domains in one deep week.
22 hrs
▼
AI in Education (EdTech)
THEORY
Personalised learning systems: knowledge graphs, spaced repetition (Anki algorithm), Bloom's taxonomy + AI, adaptive assessments.Case studies: BYJU's AI engine, Duolingo's ML model, Khan Academy's Khanmigo
PROJECT
🚀 Mini Project: Adaptive NCERT Tutor — identifies weak topics, generates personalised practice problems, tracks mastery.LLM + knowledge graph + spaced repetition. Works for Class 6–12 students.
AI in Marketing & Business
THEORY
Marketing AI: customer lifetime value (CLV) prediction, churn prevention, personalisation engines, programmatic advertising, sentiment monitoring.Tools: Meta AI (ad targeting), Google Performance Max, HubSpot AI, Salesforce Einstein
PROJECT
🚀 Mini Project: AI Content Marketing EngineCrewAI: Market Researcher + Content Strategist + SEO Specialist + Writer + Editor → full content calendar + 10 blog posts + 30 social posts → scheduled automatically
AI in Finance & FinTech
THEORY
FinTech AI: algorithmic trading, fraud detection, credit scoring, robo-advisors, KYC automation, AML (anti-money laundering).India-specific: PhonePe fraud detection, Paytm risk engine, Zerodha Kite's AI
PROJECT
🚀 Mini Project: Algorithmic Trading BotRL agent (PPO) trained on 10 years NSE data → learn buy/sell/hold policy → backtest → 23% annual return vs Nifty 50's 14%
AI in Logistics & Supply Chain
THEORY
Logistics AI: route optimisation (Vehicle Routing Problem), demand forecasting, warehouse robotics, last-mile delivery, predictive maintenance.Case: Amazon's 320+ robots in warehouses. Flipkart's AI supply chain. FedEx's SenseAware.
PROJECT
🚀 Mini Project: Smart Warehouse Management SystemComputer vision (YOLOv8) for inventory counting + RL for robot path optimisation + demand forecasting (LSTM) + Streamlit dashboard
WEEK 41
AI in Infrastructure, Smart Cities & Agriculture
Smart city systems, AI for climate, precision agriculture, energy grids, traffic management, disaster response.
20 hrs
▼
THEORY
Smart Cities: AI traffic management (SCOOT), smart energy grids (predict demand, balance load), AI water management, smart waste collection.India: Andhra Pradesh's Real-Time Governance Centre, Pune Smart City AI dashboard
THEORY
Precision Agriculture: crop disease detection from drone imagery, soil health AI, weather-driven irrigation, yield prediction, market price forecasting for farmers.India: IARI's AI crop advisory, Fasal.in sensor platform, Microsoft's FarmBeats
LAB
Build: Crop Disease Detector using CNN on PlantVillage dataset (54,306 images, 26 diseases, 14 crops).Deploy as mobile app with camera — farmer points at leaf → disease identified in 2 seconds
THEORY
AI for Climate: climate modelling (FourCastNet by NVIDIA), carbon footprint tracking, renewable energy optimisation, wildfire detection.Google DeepMind's AlphaTensor, NVIDIA's Earth-2 climate simulation
PROJECT
🚀 Project: AgriBot — Complete AI Agriculture Platform (DARPA-Level)Drone imagery (crop disease) + satellite data (soil moisture) + weather API + market price API + Hindi voice interface (Whisper + TTS) + LLM agronomist + Raspberry Pi soil sensor. The complete AI agriculture system for Indian farmers.
WEEK 42
Reinforcement Learning & Autonomous Systems
RL fundamentals, Q-Learning, DQN, PPO, RLHF, game playing AI, robotics control, self-driving systems.
22 hrs
▼
THEORY
RL fundamentals: agent, environment, state, action, reward, policy, value function. Markov Decision Process (MDP).Bellman equation: the recursive foundation of all value-based RL
THEORY
Q-Learning → Deep Q-Network (DQN) → Double DQN → Dueling DQN. Experience replay, target network.DQN beats human Atari scores: the 2013 DeepMind paper that started deep RL
THEORY
Policy gradient: REINFORCE, Actor-Critic, A3C, PPO (Proximal Policy Optimisation). SAC for continuous actions.PPO powers ChatGPT's RLHF — same algorithm, different application
IDE
OpenAI Gymnasium (Gym): environments, wrappers, custom env creation. Stable-Baselines3 for quick RL experiments.Train PPO agent on LunarLander-v2 in 30 min → achieves 250+ mean reward
LAB
Train DQN to play CartPole, MountainCar, Pong. Visualise learned policy. Ablation study on hyperparameters.Advanced: Multi-agent RL with PettingZoo — competitive and cooperative environments
PROJECT
🚀 Project: Self-Driving Car Simulation (Udacity CarSim)PPO agent learns to drive in simulation: stay on lane, avoid obstacles, obey traffic lights. Camera → CNN → RL policy → steering/throttle commands.
WEEK 43–44
MLOps — Production AI Systems at Scale
CI/CD for ML, MLflow, DVC, Kubernetes for AI, model monitoring, data drift, A/B testing ML models, cost optimisation.
26 hrs
▼
THEORY
MLOps maturity model: Level 0 (manual) → Level 1 (automated training) → Level 2 (automated ML pipeline). Why most AI fails in production.Statistics: 87% of ML models never make it to production (Gartner). MLOps solves this.
IDE
MLflow: experiment tracking, parameter logging, metrics, artefacts, model registry, model serving. Compare 50 experiments at once.Tools: MLflow 2.x, Weights & Biases (wandb), Neptune.ai, Comet.ml
IDE
DVC: Data Version Control. Track large datasets with Git. Remote storage (S3, GCS, Azure). Data pipelines.Reproduce any experiment from 6 months ago — the reproducibility crisis in ML
IDE
CI/CD for ML: GitHub Actions pipelines — auto-test models on PR, auto-train on new data, auto-deploy if metrics improve.Tools: GitHub Actions, Jenkins, ClearML, ZenML, Kubeflow Pipelines
LAB
Model monitoring: detect data drift (Evidently AI), concept drift, feature drift. Set up alerts for model degradation.Simulate: Model trained in Jan fails in July due to distribution shift — catch it early
IDE
Kubernetes for AI: deploy ML models as scalable pods. Horizontal pod autoscaler for traffic spikes. KEDA for GPU scaling.Tools: Kubernetes, Helm, Seldon Core, KServe, BentoML, Ray Serve
PROJECT
🚀 Project: Full MLOps Pipeline for a Production Recommendation SystemDVC data versioning → MLflow experiment tracking → GitHub Actions CI/CD → Docker → Kubernetes → Grafana monitoring dashboard. End-to-end production ML.
WEEK 45
AI Safety, Ethics, Alignment & Governance
AI alignment problem, Constitutional AI, bias auditing, EU AI Act, India AI Policy, responsible AI development.
18 hrs
▼
THEORY
AI alignment problem: why a superintelligent AI optimising the wrong objective could be catastrophic. Paperclip maximiser thought experiment.Anthropic's mission: beneficial AI for humanity. Constitutional AI methodology.
THEORY
Bias in AI: historical bias, representation bias, measurement bias, algorithmic bias. Famous cases: Amazon hiring AI (sexist), COMPAS (racist).Audit your own model for bias: disparate impact analysis, fairness metrics
LAB
Fairness toolkit: IBM AI Fairness 360, Google's What-If Tool, Microsoft Fairlearn. Reweigh, Calibrated Eq. Odds.Hands-on: Debias a credit scoring model that discriminates against women
THEORY
Regulation landscape: EU AI Act (risk tiers), US EO on AI, India's DPDP Act, ISO/IEC 42001 AI management standard.GDPR implications for AI: right to explanation, right to erasure, consent for ML training
THEORY
Interpretability: LIME, SHAP, Integrated Gradients, attention visualisation, mechanistic interpretability (Anthropic research).Why black-box AI is unacceptable in healthcare and law
WEEK 46
Emerging AI — World Models, Embodied AI & the Road to AGI
Sora, Gemini 1.5, Claude's capabilities, world models, embodied AI, AGI timelines, what comes next.
18 hrs
▼
THEORY
World models: AI that understands physics, causality, and can predict future states. Dreamer, JEPA (Yann LeCun's vision), Genie (Google DeepMind).Why world models are the missing piece between current AI and AGI
THEORY
Embodied AI: Figure 01 robot + ChatGPT, Google DeepMind's RT-2, Tesla Optimus, Boston Dynamics with LLM brain.The convergence: physical robots + foundation models + RL = general-purpose robots
THEORY
Multimodal frontier: GPT-4o (real-time audio+vision+text), Gemini 1.5 Pro (1M context, video understanding), Claude 3.5 (computer use).Live demo: Claude controlling a computer — the start of AI that can use any software
THEORY
AGI discussion: Definitions, timelines (OpenAI: 2025–2030, DeepMind: 2030+), AGI safety, superalignment, what AGI means for India.Read: "Situational Awareness" by Leopold Aschenbrenner — the most important AI essay of 2024
RESEARCH
Read and discuss: Key 2024–2025 AI papers — Claude's constitution, Gemini 1.5 technical report, GPT-4 technical report, Llama 3 paper.How to read AI papers: abstract → intro → results → methods. Skip the math first.
WEEK 47–48
Grand Capstone + Portfolio + Placement Preparation
Choose a Grand Capstone project, build portfolio, mock interviews, GitHub review, and Krenx internship onboarding.
40 hrs
▼
Grand Capstone Projects (Choose One)
PROJECT
🏥 HealthOS India: Complete AI healthcare platform — diagnosis + drug interaction + telemedicine + EMR + billing. Multi-agent + RAG + fine-tuned LLM + mobile app.
PROJECT
🌾 AgriOS India: End-to-end AI platform for Indian farmers — crop disease + market prices + weather + Kisan Credit scoring + Hindi voice. IoT sensors + drones + cloud.
PROJECT
🏙️ SmartCity AI: Real-time city management — traffic optimisation + crime prediction + energy management + citizen services chatbot + emergency response routing.
PROJECT
🤖 RoboLearn: AI-powered robot that teaches children — ROS2 robot + vision + NLP + adaptive curriculum + emotion recognition + parent dashboard.
PROJECT
💹 FinAI Platform: Full FinTech AI — fraud detection + credit scoring + robo-advisor + algorithmic trading + KYC automation + real-time risk monitoring.
PROJECT
🛡️ SecureAI: Cybersecurity platform — anomaly detection + threat intelligence + automated incident response + deepfake detector + social engineering defence.
Portfolio Building
LAB
GitHub portfolio: README crafting, demo GIFs, badges, project documentation, GitHub Pages portfolio site.Every project must have: demo link, tech stack, results metrics, architecture diagram
LAB
Kaggle competitions: leaderboard strategy, kernel writing, notebook upvotes. Build Kaggle reputation (Contributor → Expert).Submit to 3 Kaggle competitions during this week
LAB
LinkedIn profile optimisation for AI jobs: headline, About section, Skills, certifications, recommendations strategy.Post your grand capstone project on LinkedIn — learn to write technical posts that go viral
Interview Preparation
MOCK
Mock Technical Interview 1: ML theory — 45 mins. Explain any ML algorithm in 2 mins. Code a function from scratch.Covered: bias-variance, gradient descent, backprop, attention mechanism, RAG
MOCK
Mock Technical Interview 2: System design — Design a real-time fraud detection system for 1M transactions/day.Live coding: Python problem solving + ML problem from scratch
PANEL
Industry panel: AI engineers from top companies share interview experiences, what they look for, how to stand out.Q&A session + 1-on-1 resume review with panelists
Krenx Internship Onboarding
KRENX
Krenx Technologies Pvt. Ltd. (krenx.in) internship briefing: project allocation, team structure, tools, expectations, NDAs.3-month internship: real product development, mentored by Krenx engineers, stipend
EXAM
Advanced Diploma Final Examination: 90-minute comprehensive exam + 30-minute viva + portfolio presentation.Passing: NSQF Level 13–14 Diploma + Krenx internship offer letter
TECH STACK
70+ Tools You'll Master
Every tool used by Google, Anthropic, Amazon, and top AI startups. You'll use all of them in real projects.
Python 3.12
Core Language
NumPy
Numerical Computing
Pandas
Data Wrangling
Matplotlib
Visualisation
Seaborn
Statistical Viz
Plotly
Interactive Charts
Scikit-Learn
Machine Learning
PyTorch
Deep Learning
TensorFlow 2.x
Deep Learning
Keras
DL High-Level API
HuggingFace
LLMs & Transformers
LangChain
LLM Applications
LangGraph
AI Workflows
CrewAI
Multi-Agent
AutoGen
Multi-Agent
LlamaIndex
RAG Framework
OpenAI API
GPT-4o, DALL·E
Claude API
Anthropic
Gemini API
Google AI
Ollama
Local LLMs
Pinecone
Vector Database
ChromaDB
Vector Database
FAISS
Vector Search
Stable Diffusion
Image Generation
OpenCV
Computer Vision
YOLOv8/v11
Object Detection
MLflow
Experiment Tracking
Docker
Containerisation
Kubernetes
Orchestration
FastAPI
API Development
Streamlit
ML Web Apps
Google Cloud AI
Cloud ML
AWS SageMaker
Cloud ML
ROS2
Robotics
Gazebo
Robot Simulation
Raspberry Pi
Edge AI / IoT
TFLite
Edge Inference
Unsloth
LLM Fine-Tuning
PEFT / LoRA
Fine-Tuning
SHAP / LIME
Explainability
GitHub Copilot
AI Coding
Gym / SB3
Reinforcement RL
Whisper
Speech AI
spaCy / NLTK
NLP
Evidently AI
Model Monitoring
XGBoost
Gradient Boosting
LightGBM
Gradient Boosting
Optuna
Hyperparameter Opt
Weights & Biases
Experiment Tracking
DVC
Data Versioning
PostgreSQL
Database
REAL-WORLD APPLICATIONS
AI Across Every Domain
Learn to apply AI to transform every major industry. Each domain module includes tools, real datasets, industry case studies, and a hands-on project.
Healthcare & Medicine
Medical imaging, drug discovery, genomics, clinical NLP, predictive health.
- X-ray/MRI/CT AI diagnosis
- AlphaFold protein structure
- Clinical note NLP (EHR)
- Epidemic prediction models
- Telemedicine AI triage
- Surgical robot vision
Defence & National Security
Autonomous systems, surveillance AI, threat detection, electronic warfare.
- Drone swarm coordination
- Missile guidance AI
- Battlefield surveillance
- Counter-drone systems
- Cyber threat AI
- OSINT and deepfake detection
Agriculture & Food Security
Precision farming, crop disease, yield prediction, water management.
- Drone crop disease detection
- Satellite soil analysis
- Market price forecasting
- Automated irrigation AI
- Pest prediction models
- Supply chain optimisation
Finance & FinTech
Algorithmic trading, fraud detection, robo-advisors, credit scoring.
- Real-time fraud detection
- Algorithmic trading (RL)
- AI credit underwriting
- Robo-advisory platforms
- Anti-money laundering AI
- Automated KYC/KYB
Education & EdTech
Personalised learning, intelligent tutoring, automated assessment.
- Adaptive learning paths
- AI writing evaluation
- Plagiarism detection
- Student dropout prediction
- Personalised NCERT tutor
- AI exam generation
Manufacturing & Industry 4.0
Predictive maintenance, quality control, process optimisation, cobots.
- Predictive maintenance (vibration AI)
- Visual quality inspection
- Digital twins for factories
- Energy consumption AI
- Supply chain resilience
- Collaborative robot vision
Logistics & Supply Chain
Route optimisation, warehouse robotics, demand forecasting, last-mile AI.
- Vehicle routing optimisation
- Warehouse pick-and-place robots
- Demand forecasting (LSTM)
- Last-mile delivery AI
- Port container tracking
- Cold chain monitoring IoT
Marketing & Media
Personalisation, content generation, sentiment monitoring, ad targeting.
- Customer lifetime value AI
- AI content generation (CrewAI)
- Social media sentiment
- Programmatic advertising AI
- Churn prediction models
- Influencer scoring AI
Smart Cities & Infrastructure
Traffic management, energy grids, emergency response, citizen services.
- Adaptive traffic signal AI
- Smart energy grid optimisation
- Crime prediction and prevention
- Emergency response routing
- AI citizen services portal
- Smart water management
Climate & Environment
Climate modelling, carbon tracking, wildfire detection, renewable energy.
- Weather prediction (FourCastNet)
- Carbon footprint tracking
- Wildfire early detection
- Solar/wind output prediction
- Flood risk mapping AI
- Species identification (vision)
Legal & Governance
Legal document analysis, contract review, case prediction, regulatory AI.
- IPC/CrPC document search AI
- Contract clause extraction
- Case outcome prediction
- Regulatory compliance AI
- Patent similarity search
- Judicial backlog analysis
Scientific Research
Drug discovery, materials science, particle physics, genomics, climate science.
- AlphaFold protein structure
- Materials property prediction
- Genomics variant calling
- Astronomy: exoplanet detection
- CERN data analysis AI
- Literature review automation
KRENX TECHNOLOGIES
krenx.in · Private Limited · Internship Partner
3-Month Industry Internship Program
Exclusively for Remesys Advanced Diploma graduates. Work on real products at Krenx Technologies, mentored by senior AI engineers. Build your professional portfolio, earn a stipend, and convert to a full-time role.
Duration
3 months (12 weeks), Monday to Friday, 9AM–6PM. Remote/hybrid options available.
Stipend
Monthly stipend provided. Amount based on project and performance. Not unpaid!
Real Projects
Work on live Krenx products used by real clients. Not dummy projects.
Certificate
Krenx internship completion certificate + recommendation letter + LinkedIn recommendation.
INTERNSHIP TRACKS
Choose Your Track
TRACK A
AI Product Engineering
Build LLM-powered products. RAG systems, chatbots, agentic workflows for real Krenx clients. Stack: LangChain, FastAPI, React, Docker, AWS.
TRACK B
Computer Vision & Edge AI
Deploy vision models to edge devices for industrial/security clients. Stack: YOLOv8, TFLite, OpenCV, Raspberry Pi, ROS2.
TRACK C
Data Science & Analytics
End-to-end data pipelines, dashboards, ML models for Krenx business clients. Stack: Pandas, scikit-learn, Streamlit, PostgreSQL, Grafana.
TRACK D
IoT & Robotics Systems
Build IoT+AI systems for smart factory / smart city clients. Stack: Raspberry Pi, MQTT, ROS2, TensorFlow Lite, Gazebo.
INTERNSHIP TIMELINE
12-Week Journey
WEEK 1–2 · ONBOARDING
Team integration, codebase review, tool setup
Meet your team, get access to systems, understand the product, set goals with your mentor.
WEEK 3–6 · FOUNDATION
First feature contribution
Implement a defined feature under mentorship. Daily standups, code reviews, Git workflow.
WEEK 7–10 · OWNERSHIP
Lead a module independently
Own a feature/module end-to-end. Present weekly progress. Client demo preparation.
WEEK 11–12 · DELIVERY
Final project presentation + evaluation
Demo to Krenx management + client. Performance review. Return offer discussion. Graduation!
PORTFOLIO BUILDING
50+ Real-World Projects
Every project is deployable, shareable, and based on actual industry use-cases. Your portfolio will speak louder than any resume.
🏥
INTERMEDIATE · NSQF 12–13 · TENSORFLOW
Medical X-Ray AI Diagnostic System
DenseNet-121 detects 14 lung pathologies from chest X-rays. Grad-CAM heatmap shows "where AI is looking." Clinical report generated by LLM. Achieves 97.3% AUC on NIH dataset.
🌾
DIPLOMA · NSQF 13–14 · MULTI-AGENT
AgriBot — Complete AI Farming Platform
Drone imagery + satellite data + IoT soil sensors + weather API + Hindi voice interface + LLM agronomist + market price alert. The most complete AI agriculture system for Indian farmers.
💬
ADVANCED · NSQF 13 · RAG + LANGCHAIN
Indian Legal AI Assistant
RAG over 5000+ IPC sections + Supreme Court judgements. Ask legal questions in Hindi or English. Get answers with exact section references. Multi-source hybrid search. Deployed on WhatsApp.
🤖
DIPLOMA · NSQF 14 · ROS2 + AI
Autonomous Medical Delivery Robot
ROS2 + Nav2 + SLAM + YOLOv8. Navigates hospital corridors autonomously, identifies patient rooms, delivers medication, avoids moving obstacles. Full Gazebo simulation with RViz2 visualisation.
🛡️
DIPLOMA · NSQF 14 · CYBERSECURITY AI
Real-Time Cyber Threat Detection Platform
Isolation Forest + Autoencoder detect network anomalies in real-time. LLM explains each alert in plain English. Deepfake detector for phishing defence. Automated incident response playbooks.
💹
ADVANCED · NSQF 13 · RL TRADING
Algorithmic Trading RL Agent (NSE/BSE)
PPO reinforcement learning agent trained on 10 years of NIFTY 50 data. Learns buy/sell/hold policy. Achieves 23% annual return vs 14% benchmark. Live paper trading with Zerodha Kite API.
ENROLL TODAY
Transparent Pricing
No hidden charges. 40% spot discount on all courses. Limited seats. All prices include LMS, T-Shirt, ID Card, 24/7 AI support, parent dashboard, and cloud IDE.
40% OFF
3 MONTHS · NSQF LEVEL 12
AI BASIC
🇮🇳 NSQF Level 12 · NiELIT Certified
₹9,999
₹5,999
+ ₹1,000 one-time admission fee
🔥 SAVE ₹4,000 — SPOT DISCOUNT
✓ Python, NumPy, Pandas, Matplotlib
✓ Scikit-Learn Machine Learning
✓ Data Science pipeline
✓ 5 Real Projects (deployed)
✓ NSQF Level 12 Certificate
✓ LMS + T-Shirt + ID Card
✓ 24/7 AI Doubt Support
✓ Parent Dashboard
✓ Cloud Practice IDE
✓ Scikit-Learn Machine Learning
✓ Data Science pipeline
✓ 5 Real Projects (deployed)
✓ NSQF Level 12 Certificate
✓ LMS + T-Shirt + ID Card
✓ 24/7 AI Doubt Support
✓ Parent Dashboard
✓ Cloud Practice IDE
40% OFF
6 MONTHS · NSQF LEVEL 12–13
AI INTERMEDIATE
🇮🇳 NSQF Level 12–13 · NiELIT Certified
₹19,999
₹11,999
+ ₹1,000 one-time admission fee
🔥 SAVE ₹8,000 — SPOT DISCOUNT
✓ All Basic content
✓ Deep Learning (TF + PyTorch)
✓ CNNs + Object Detection (YOLO)
✓ Transformers + NLP + RAG
✓ 12 Real Projects (deployed)
✓ Internship Support
✓ NSQF Level 12–13 Certificate
✓ All benefits from Basic tier
✓ Deep Learning (TF + PyTorch)
✓ CNNs + Object Detection (YOLO)
✓ Transformers + NLP + RAG
✓ 12 Real Projects (deployed)
✓ Internship Support
✓ NSQF Level 12–13 Certificate
✓ All benefits from Basic tier
40% OFF
9 MONTHS · NSQF LEVEL 13
AI ADVANCED
🇮🇳 NSQF Level 13 · NiELIT + IEEE Aligned
₹39,999
₹23,999
+ ₹1,000 one-time admission fee
🔥 SAVE ₹16,000 — SPOT DISCOUNT
⭐ BEST VALUE — MOST POPULAR
✓ All Intermediate content
✓ Gen AI · LLMs · Fine-Tuning · RAG
✓ Agentic AI · Multi-Agent Systems
✓ IoT + Robotics (ROS2, Raspberry Pi)
✓ AI in 12+ Domains
✓ 20+ Real Industry Projects
✓ Placement Assistance
✓ NSQF Level 13 Certificate
✓ Gen AI · LLMs · Fine-Tuning · RAG
✓ Agentic AI · Multi-Agent Systems
✓ IoT + Robotics (ROS2, Raspberry Pi)
✓ AI in 12+ Domains
✓ 20+ Real Industry Projects
✓ Placement Assistance
✓ NSQF Level 13 Certificate
40% OFF
12 MONTHS · NSQF LEVEL 13–14
ADV. DIPLOMA
🇮🇳 NSQF Level 13–14 · Research Grade
₹79,999
₹47,999
+ ₹1,000 one-time admission fee
🔥 SAVE ₹32,000 — SPOT DISCOUNT
✓ COMPLETE AI MASTERY
✓ All Advanced content
✓ MLOps + AI Safety + Research
✓ Emerging AI: World Models, AGI
✓ 30+ Industry Projects
✓ Grand Capstone Project
✓ 100% Placement Drive
✓ 3-Month Krenx Technologies Internship
✓ NSQF Level 13–14 Diploma
✓ All Advanced content
✓ MLOps + AI Safety + Research
✓ Emerging AI: World Models, AGI
✓ 30+ Industry Projects
✓ Grand Capstone Project
✓ 100% Placement Drive
✓ 3-Month Krenx Technologies Internship
✓ NSQF Level 13–14 Diploma
🇮🇳 GOVT. OF INDIA APPROVED
All Remesys AI courses are aligned with the National Skills Qualifications Framework (NSQF) at Levels 12–14, recognised by the Ministry of Education, Government of India. Aligned with NiELIT standards, National Education Policy (NEP 2020), and international standards including ISO/IEC 42001 (AI Management) and IEEE AI ethics guidelines.
NSQF LEVELS
NSQF 12 · Basic (3 mo)
NSQF 12–13 · Intermediate (6 mo)
NSQF 13 · Advanced (9 mo)
NSQF 13–14 · Diploma (12 mo)
REMESYS
India's Premium AI Learning Institute
www.remesys.in
Internship Partner: Krenx Technologies Pvt. Ltd. · krenx.in
📱 WhatsApp Enquiry
🎓 NSQF Level 12–14
🇮🇳 Govt. of India Approved
🔥 40% Spot Discount Active