Learn In தமிழ் 
4.8K+ students have already registered
Why choose this course?
Prestigious certification from Vishlesan i-Hub, IIT Patna
Real-world capstone projects designed to solve practical challenges
Delivery in Tamil, learn from experts who speak your language
Placement assistance for 1 year, plus resume review and interview coaching sessions facilitated by Masai. (Eligibility: 7+ CGPA and 65% attendance)
Expert faculty and industry insights
*Certificate Eligibility: Minimum 65% overall attendance and 30% marks in cumulative evaluation score
Career Outcomes
AI Solutions Engineer
₹5 - ₹47 LPA
Deploy AI systems, automate workflows & integrate business AI tools.
Source: GlassdoorSoftware Dev Engineer
₹12 - ₹37.4 LPA
Scalable product engineering & AI-integrated software systems.
Source: 6figrML Engineer
₹10 - ₹35 LPA
Build & deploy production ML models across fintech, e-commerce & health-tech.
Source: Industry avg3× Avg salary vs non-tech peers after just 1 year
70% Of AI engineer roles require Python + ML skills
180% Growth in AI jobs in Tamil Nadu in 2026
NOW Zoho, Freshworks & TCS actively hiring AI engineers from Tamil Nadu
What will you learn?
Go from zero to job-ready in AI/ML with beginner-friendly, industry-aligned curriculum from I-Hub IIT Patna, taught in Tamil — covering Python, data analysis, machine learning, and generative AI with hands-on projects. Built for beginners, designed to get you building real things fast.
Meet our Instructor Team
Abhinandan S P
Asst. Professor at IIT Palakkad
Aswath Venkatesan
Founder, KNOT Academy | Ex Walmart
Aswath Rao V S
AI / ML Engineer at Ford Motors, Ex- PayPal and Pine Labs
Abhinandan S P
Asst. Professor at IIT Palakkad
Aswath Venkatesan
Founder, KNOT Academy | Ex Walmart
Aswath Rao V S
AI / ML Engineer at Ford Motors, Ex- PayPal and Pine Labs
*Note: Instructors teaching your batch may change based on specific module expertise, or circumstances such as personal emergencies, unavailability etc.
Course Details
Foundations of Data
- AI vs ML vs GenAI landscape Git/GitHub/CLI VS Code/Colab environments & secrets
- Python for Data (types, flow, functions, files) · notebooks discipline
- NumPy + pandas (indexing, joins/merge, groupby, missing values)
- SQL lite (SELECT/WHERE/GROUP BY + one JOIN) · spreadsheets (pivots, XLOOKUP) · pandas ↔ SQL mental model
- Plotting & storytelling (matplotlib/plotly) · EDA checklist
- APIs (requests, JSON) + basic ethics/ToS · ingest → clean → EDA snack app (optional Streamlit)
Mini Projects Unlocked
CryptoWatch — Real-Time Market Sentiment Tracker
FoodSafety Insights — Restaurant Inspection Intelligence
TrendPulse — What's Actually Trending Right Now
Classical ML
- ML workflow & habits: problem framing, train/validation/test split, baselines, cross‑validation, metric tables
- Data prep: missing values, duplicates, categorical encoding, scaling, leakage guard, simple feature engineering (dates, counts), class imbalance basics
- Regression: Linear Regression + regularization; metrics MAE/RMSE/R²; quick error analysis
- Classification: Logistic Regression; Trees & Random Forests; metrics accuracy, precision/recall, F1, ROC‑AUC; threshold tuning
- Model comparison & selection: choose by metric + simplicity
Mini Projects Unlocked
ChurnGuard — Predicting Customer Exit Before It Happens
LoanLens — Automated Credit Risk Assessment
RentRight — Predicting Fair Apartment Prices
GenAI & Agents
- LLM fundamentals: tokens, context windows, temperature/top‑p, cost/latency basics
- Prompting patterns: role, few‑shot, CoT; tiny prompt‑eval set
- Agents 101 (basics): single agent, 1–2 tools, JSON schemas, retries/backoff, hard stop; mini‑build (CSV Analyst or FAQ Helpdesk)
- Function/Tool calling: structured outputs (JSON/validation), idempotent tool design
- Shipping Sprint (FastAPI): wrap the agent as an API ( /chat ) with Pydantic schemas, env/secrets, logging + request IDs; deploy to Render or HF Spaces; run a 10‑prompt live eval
- RAG 101: embeddings, chunking, retrievers, vector stores; RAG eval (answer correctness & context recall), semantic search
- Agents + RAG: reuse the Agents 101 loop with a retriever tool; same eval harness
- Orchestration (intro): LangGraph basics—nodes, transitions, retries; ship a small agent flow
Mini Projects Unlocked
DataGenie — Natural Language Data Analyst
SupportAI — Intelligent FAQ Helpdesk Agent
ResearchPilot — Automated Market Research Agent
Multi-Model Answerboard (AI Fiesta-style)
One chat, multiple model answers side-by-side with cost/speed badges; toggle RAG for cited answers and add smart routing + A/B prompt
English & Soft Skills Training
Everyday English for Self-Expression
Learn how to talk about yourself comfortably and confidently in common situations.
Building Clear and Correct Sentences
Strengthen sentence formation so your English sounds natural and complete.
Speaking with Impact and Awareness
Learn to organise thoughts, speak smoothly, and communicate with confidence.
Professional Communication & Soft Skills
Apply spoken English skills to professional and workplace-like situations.
Please Note :- "Due to the evolving nature of the industry expectations and partner institute feedback, some syllabus aspects may change. Any updates will be communicated during the Inauguration Session(s) or at the start of the relevant module"
Capstone Project
Nykaa Review Radar
Nykaa sells thousands of beauty products, and every month someone decides which to promote, discount or drop on pure instinct. Nobody has time to read a hundred thousand reviews, so by the time a rating visibly collapses it is too late to act.
What you'll build:
An early warning system that cleans the full review corpus into SQL and trains a model to flag products heading for a bad rating before it happens. On top sits a chat agent with a database tool, a model tool and retrieval over review text, shipped as a FastAPI service on Hugging Face Spaces.
Final Outcome
A monthly action list ranking the top 200 products as promote, hold, discount or remove. Each row carries a risk score and the real customer quotes behind it, plus a live demo link you can share.
Zerodha Market Pulse
An investor sees a stock down four percent and cannot tell if it is normal noise or the start of trouble. Finding out means reading years of filings and a week of news, so most people either panic or ignore it.
What you'll build:
A Nifty 50 assistant analyzes five years of price history to classify stocks as calm, normal, or turbulent for the upcoming month, focusing on risk conditions. It uses price and document retrieval tools, orchestrated in LangGraph and deployed on Render, explaining each call with the source.
Final Outcome
A weekly watchlist tagging every Nifty 50 stock with its risk regime and a confidence score. Each call comes with the specific news or disclosure driving it, so it can be checked rather than trusted blindly.