Batch 1 (Nov 2026 Intake) — Starts Sat, 14 Nov 2026

What you'll learn

1. 1. Week 1: Gen AI & LLM Fundamentals + SetupHow LLMs work: tokens, context window, temperature, top-p · Model landscape: Gemini, GPT, Claude, Llama/Gemma (open vs closed) · Prompting basics; where LLMs fail (hallucination, bias) · Dev setup: VS Code, Python venv, Git & GitHub · Git workflow like a team: branches, pull requests, code review and .gitignore for secrets
2. 2. Week 2: Advanced Prompting & EvaluationFew-shot, role, chain-of-thought, prompt chaining, ReAct preview · Structured output (JSON schema), system instructions · Prompt evaluation: test sets, rubrics, LLM-as-judge · Career corner: how recruiters screen AI resumes
3. 3. Week 3: Python for AI + APIsBasic Python fast-track (pre-recorded Fast-Start Setup Pack + this week's lab clinic): syntax & indentation, variables and data types, if/else and loops, functions, lists & dictionaries, reading/writing files and JSON, modules, pip and virtual environments (venv), calling a REST API with requests, and your first simple LLM call with the Gemini Python SDK · Python refresher: functions, dicts, files, exceptions · HTTP, REST, requests, JSON · API keys and secrets: .env, never commit keys, rate limits · Calling the Gemini API (Python SDK), streaming responses · SQL essentials for AI apps: SELECT, JOIN, GROUP BY with SQLite/Postgres, and safe text-to-SQL (read-only users, query validation) · Writing simple pytest unit tests for your functions
4. 4. Week 4: Your First LLM App + DeploymentStreamlit/Gradio UIs · Cost and latency basics; caching · Deploying to Streamlit Community Cloud / Hugging Face Spaces · Writing a great README + 60-second demo video
5. 5. Week 5: Embeddings, Vector DBs & Semantic SearchEmbeddings and similarity · Vector databases: Chroma, FAISS, pgvector (Supabase) · Chunking strategies, metadata filters · Semantic vs keyword vs hybrid search
6. 6. Week 6: RAG Pipeline End-to-EndRetrieve, augment, generate; grounding and citations · PDF parsing, OCR basics, multilingual (Telugu/Hindi) documents · Prompt templates for RAG; 'I don't know' behaviour · Frameworks: LangChain/LlamaIndex vs plain Python
7. 7. Week 7: Evaluating & Hardening RAGRAG evaluation: faithfulness, relevance (RAGAS-style metrics) · Reducing hallucinations; reranking · Guardrails, prompt injection, PII handling · Cost/latency optimization · OWASP Top 10 for LLM Applications: prompt injection, insecure output handling, excessive agency
8. 8. Week 8: Backend & Production DeploymentFastAPI backend, auth basics · Docker basics · Deploy to Render/Railway/Google Cloud Run (free tiers where available) · Logging and monitoring · CI/CD with GitHub Actions: lint, tests and Docker build on every push · Security basics: secret management, auth, rate limiting and least-privilege API keys · Same app on other clouds: how the stack maps to Azure (Azure OpenAI, Container Apps) and AWS (Bedrock, App Runner/ECS)
9. 9. Week 9: Function Calling & AgentsTool use / function calling · ReAct loop, planning, memory · Frameworks: LangGraph or Google Agent Development Kit (ADK) · When NOT to use agents
10. 10. Week 10: Multi-step Agents, MCP & Responsible AI + Capstone KickoffMulti-agent patterns, human-in-the-loop · Model Context Protocol (MCP) intro · Responsible AI: bias, privacy, licensing, safety · Capstone scoping: problem, users, architecture, success metric · Build a small MCP server that exposes your RAG search as a tool · Responsible AI and India's DPDP Act 2023 basics: consent, data minimisation, deleting user data
11. 11. Week 11: AI for Careers + Capstone BuildAI-assisted resume and LinkedIn, done honestly (no fabricated experience) · GitHub profile README, pinned repos, portfolio site · Mock technical interview: LLM/RAG system-design questions; HR round · Freelancing/internship hunting: Internshala, LinkedIn, Upwork basics
12. 12. Week 12: Polish & DEMO DAYTesting, UX polish, README, architecture diagram, 2-minute demo video · Pitching your project in interviews

Final project demo: Portfolio-Grade Deployed Gen AI App. Solve a real problem with a RAG and/or agent app that is deployed at a public URL. Examples: AI engineering interview-prep interview coach grounded on company JDs; Telugu/English farmer advisory bot grounded on government scheme PDFs; enterprise knowledge-base assistant with citations; small-business WhatsApp/web order assistant; AI video-script generator. Must include: a public deployed URL, a GitHub repo with a clean README (problem, architecture diagram, setup, eval results), a 2-minute demo video, basic evaluation, and no secrets in the repo.