Enrolling now
Generative AI & Agentic AI Professional Program
12 weeks · live classes · LLMs, RAG, fine-tuning, agents & MCP
A 12-week live program that takes you from LLM basics and prompt engineering to production RAG, fine-tuning, AI agents, multi-agent systems and MCP, and ends with a deployed agentic AI capstone.
LLM basics, prompt engineering-ல இருந்து production RAG, fine-tuning, AI agents, multi-agent systems, MCP வரைக்கும் 12 வாரத்துல live-ஆ கத்துக்கலாம். கடைசில ஒரு agentic AI capstone-ஐ cloud-ல deploy பண்ணுவோம்.
- 12 weeksfrom the first class to the capstone
- 12 modules167 topics, every one in the syllabus below
- Live in Tamilwith recordings and notes in English
- Hands-onevery topic tried on your own machine
What you will work with
Python
LangChain
FastAPI
Docker
GitHub Actions
AWSOpenAIClaudeGeminiOllamaHugging FaceLlamaIndexLangGraphCrewAIAutoGenMCPChromaPineconeQdrantFAISSRAGASQLoRAChainlitLangfuse
By the end, you can
- Explain how LLMs work, from tokens and attention to context windows and cost
- Apply prompt engineering techniques and get reliable structured output
- Build applications on OpenAI, Claude, Gemini and local models with LangChain
- Build production RAG with hybrid search, reranking and RAGAS evaluation
- Use corrective, adaptive and agentic RAG, and query both documents and databases
- Fine-tune an open model with LoRA and QLoRA and measure the result
- Build agents from scratch and with LangGraph, CrewAI and AutoGen
- Build MCP servers and add memory and guardrails to agents
- Deploy, monitor and operate AI systems with Docker, CI/CD and tracing
What you get
- Live classesTwo live lectures a week, taught in Tamil and English, with time to ask questions as they come up. Every class is recorded, so a missed session is never lost.
- Weekly hands-on labA guided lab every week, so you learn by building and not just by watching.
- Doubt clearingDoubt-clearing sessions every week, plus help when you are stuck on a lab or a project.
- Reading materials and notebooksStudy materials, notebooks and project starter templates in the portal. Read at your own pace and come back to revise any time.
- Mini projects every weekA small project at the end of most weeks, so you finish with a portfolio of working AI systems.
- Capstone with reviewOne end-to-end capstone, planned, reviewed and deployed with you, and presented at the end.
- Career supportResume and GitHub portfolio review, mock technical and system design interviews, and interview preparation.
- CertificateA completion certificate once your capstone is accepted.
The full syllabus
Every topic below is something you run on your own machine, in a lab we give you, before you see it in the cloud. Open a module to see each topic in it.
Module 1 · Foundations & LLM Essentials
Week 1 · Generative AI and your development environment15 topics
Lab · set up a complete AI development environment.
The Generative AI landscape
- What Generative AI is, what it can do and where it fails
- From rule-based systems to large language models
- Text, image, audio, video and code generation
- Industry use cases, and the opportunity in India
- Responsible AI and ethics
Your AI development environment
- Python essentials for GenAI work
- VS Code with AI extensions, Jupyter and Google Colab
- Git and GitHub for AI projects
- Keeping API keys and environment variables safe
- Cloud basics on AWS and Google Cloud
Understanding LLMs
- Transformers and attention, at a high level
- GPT, Claude, Gemini, LLaMA and Mistral
- Open-source or proprietary, and when to use which
- Tokens, context windows and model parameters
- Token economics and keeping cost down
Week 2 · Prompt engineering16 topics
Lab · a prompt playground and a reusable template library. Mini project · an AI content generation system.
Fundamentals
- Anatomy of a good prompt
- System, user and assistant messages
- Temperature, top-p and other generation settings
- Managing the context window, and common pitfalls
Core techniques
- Zero-shot, one-shot and few-shot prompting
- Role-based prompting
- Structured output as JSON, XML and Markdown
- Prompt templates and reuse
Advanced strategies
- Chain of thought and tree of thought
- Self-consistency and ensembles
- The ReAct pattern
- Prompt chaining and meta-prompting
Best practices
- Refining prompts step by step, and A/B testing them
- Reducing hallucinations and checking facts
- Spotting and reducing bias
- Documenting and versioning prompts
Week 3 · LLM APIs and your first LLM application17 topics
Lab · a multi-model chatbot with memory and a document Q&A. Mini project · a support chatbot with escalation.
OpenAI API
- Authentication and request structure
- Chat completions, roles and streaming
- Function calling and structured outputs
- Vision, Whisper and text to speech
- Rate limits and error handling
Other LLM APIs
- The Claude Messages API
- Gemini and multimodal input
- Groq, Together AI and Replicate
- Hugging Face Inference API and Transformers
- Running models locally with Ollama
LangChain
- Models, prompts and output parsers
- LangChain Expression Language (LCEL) and chains
- Memory for conversations
- Document loaders and text splitters
Building the application
- Architecture patterns for LLM apps
- Chatbots with memory and streaming replies
- Fallbacks, logging and monitoring basics
Week 4 · RAG and vector databases15 topics
Lab · Chroma and Pinecone, an embeddings pipeline and semantic search. Module 1 project · chat with your documents.
Retrieval-augmented generation
- Why plain LLMs are not enough
- The RAG architecture and pipeline
- RAG, fine-tuning or prompt engineering, and when to use each
- Measuring RAG quality
Embeddings and semantic search
- What an embedding is, with the intuition behind it
- OpenAI, Cohere and Sentence Transformers embeddings
- Dimensions and their trade-offs
- Cosine, Euclidean and dot product similarity
Vector databases
- Why vectors need their own storage
- Pinecone, Chroma, Weaviate, Qdrant and FAISS compared
- HNSW, IVF and PQ indexes
- Metadata filters and hybrid search
A basic RAG pipeline
- Loading and cleaning documents
- Fixed-size, recursive and semantic chunking
- Retrieval, reranking and putting context into the prompt
Module 2 · Advanced LLM & Production RAG
Week 5 · Production-ready RAG13 topics
Lab · compare chunking strategies, hybrid search, RAGAS evaluation. Mini project · an enterprise knowledge base.
Advanced chunking
- Semantic and document-aware chunking
- Parent-child chunks and overlapping windows
- Chunking PDF, HTML and code
Advanced retrieval
- Hybrid keyword and semantic search
- Query expansion, multi-query and HyDE
- Maximal marginal relevance
- Contextual compression and filtering
LlamaIndex
- Ingestion and indexing
- Query engines, retrievers and response synthesis
- Structured data and multi-document queries
Evaluation and tuning
- Faithfulness, relevance and context precision
- Evaluating with RAGAS
- Debugging retrieval failures and A/B testing configurations
Week 6 · Advanced RAG architectures13 topics
Lab · corrective RAG, text-to-SQL, multi-modal documents. Mini project · a financial analyst RAG system.
Agentic RAG patterns
- Router RAG
- Corrective, adaptive and self-RAG
- Multi-hop RAG for complex questions
Multi-modal RAG
- Vision-language models for documents
- Tables, charts and diagrams
- Audio and video content
RAG over structured data
- Text-to-SQL
- Knowledge graphs and Graph RAG
- Combining structured and unstructured data
Running RAG in production
- Scaling, caching and cost control
- Security and access control
- Keeping data fresh with incremental updates
- Monitoring and observability
Week 7 · Fine-tuning LLMs13 topics
Lab · build an instruction dataset and fine-tune a 7B model with QLoRA. Mini project · a Tamil-English bilingual model.
When and how to fine-tune
- Fine-tuning, RAG or prompting
- Full fine-tuning, LoRA and QLoRA
- Domain adaptation and the cost-benefit question
Preparing the dataset
- Collecting and cleaning data
- Alpaca, ShareGPT and OASST formats
- Synthetic data and augmentation
- Train, validation and test splits
Parameter-efficient fine-tuning
- LoRA and QLoRA explained
- Adapters, prefix tuning and prompt tuning
- Choosing the right method
The training workflow
- Hugging Face Transformers, PEFT and Unsloth
- Choosing hyperparameters and watching training
- Evaluating, merging and deploying the model
Week 8 · LLM application UI, APIs and testing13 topics
Lab · a Chainlit chat UI, a FastAPI backend and a test suite. Module 2 project · an enterprise RAG application.
Chat interfaces
- Streamlit, Gradio and Chainlit
- Streaming replies, message history and state
APIs with FastAPI
- REST APIs for LLM services
- Async inference
- Validation, errors and OpenAPI docs
- Rate limiting and authentication
Architecture patterns
- Microservices and event-driven design
- Queues for long-running tasks
- Caching with Redis and in memory
- Database integration
Testing
- Unit and integration tests for LLM apps
- Prompt regression tests
- Load testing and continuous integration
Module 3 · Agentic AI & Production Deployment
Week 9 · Foundations of AI agents14 topics
Lab · the four agentic patterns and a ReAct agent from first principles. Mini project · a research assistant agent.
What an agent is
- Agents, chatbots and workflow automation compared
- Perception, reasoning and action
- Agent architectures and real-world uses
Agentic design patterns
- Reflection
- Tool use
- Planning
- Multi-agent collaboration, and when to use which
Tool calling
- OpenAI function calling and Claude tool use
- Building custom tools such as search, a calculator and code execution
- Choosing tools, handling failures and keeping tool use safe
An agent from scratch
- The agent loop and the observe, think, act cycle
- Short-term and long-term memory
- State management
- A ReAct agent built by hand
Week 10 · Agent frameworks and orchestration11 topics
Lab · LangGraph, CrewAI and AutoGen side by side. Mini project · a software development agent crew.
LangGraph
- State machines, nodes, edges and conditional routing
- Persistence and checkpoints
- Human-in-the-loop
CrewAI
- Agents with roles and goals
- Tasks and delegation
- Sequential and hierarchical crews
AutoGen
- Conversational agents and group chat
- Code execution agents
- AutoGen Studio
Choosing a framework
- Selection criteria, performance and scale
- LangGraph, CrewAI or AutoGen, and combining them
Week 11 · MCP, multi-agent systems, memory and guardrails14 topics
Lab · build and deploy MCP servers, memory with Mem0, guardrails. Mini project · a customer service agent system.
Model Context Protocol (MCP)
- MCP architecture, servers and clients
- Resources, tools and prompts
- Building a custom MCP server
- Connecting MCP servers to Claude Desktop
Multi-agent architectures
- Supervisor and worker agents
- Hierarchies and swarms
- How agents communicate, vote and resolve conflicts
Memory and knowledge
- Long-term memory with Mem0
- Knowledge graphs for agents
- Storing and retrieving memory efficiently
Safety and guardrails
- Prompt injection and how to defend against it
- Validating and filtering output
- PII detection and resource limits
- NeMo Guardrails and red teaming agents
Week 12 · Production deployment and capstone13 topics
Lab · monitoring with LangSmith, Docker deployment to the cloud, a CI/CD pipeline. Capstone presentation.
LLMOps
- Tracing with LangSmith and Langfuse
- Cost tracking and A/B testing in production
- Debugging and handling incidents
Deployment
- Docker and cloud deployment on AWS, GCP or Azure
- Serverless and Kubernetes options
- CI/CD with GitHub Actions
- Blue-green and canary releases
Production practices
- Security hardening and secrets management
- API versioning, load balancing and auto-scaling
- Disaster recovery and data privacy
Capstone
- Planning and architecture design
- Build sprints and code review
- Deployment, documentation and the final presentation
What you will build
- Mini project · an AI content generation system with several prompt strategies
- Mini project · a customer support chatbot with FAQ handling and escalation
- Module 1 project · a chat-with-your-documents app that answers questions from uploaded PDFs
- Mini project · an enterprise knowledge base RAG system with evaluation metrics
- Mini project · a financial analyst RAG system that queries documents and databases
- Mini project · a Tamil-English bilingual instruction-following model, fine-tuned with QLoRA
- Module 2 project · an enterprise RAG application with a UI, an API and an evaluation dashboard
- Mini project · a research assistant agent that searches, summarises and synthesises
- Mini project · a software development crew with product manager, developer and tester agents
- Mini project · a customer service agent system with memory, escalation and guardrails
- Capstone · a production agentic AI system (business analyst, shopping assistant, clinical documentation, AI tutor or your own approved idea), deployed with a live URL, architecture document, API docs and a demo video
Who it is for
- Engineering students preparing for AI careers
- Working professionals who want to upskill or move into AI roles
- Career changers entering technology
- Startup founders building AI products
- Teams that need to build AI capabilities
- You need basic Python. No prior AI or machine learning experience is needed
What our students say
In their own words, from students of our live GoDisai batches.
Concepts are explained in the simplest way to understand. Doubts are clearly explained on the spot until everyone understands the concept — very useful for preparing for interviews.
If you want to change your role with the current trend, he is the best — and you can learn in your mother tongue. I can now change my role and attend interviews confidently.
The way you teach things from absolute scratch and not from an abstraction level really made me understand a lot of technical topics with much more depth. “I am able to explain why and how things work and know how to do it” — that would be the takeaway from the sessions.
The hands-on examples and doubt-clearing sessions helped me understand how to apply these concepts in my own projects.
Clear explanation with simple examples and a practical approach to real-time issues. You can learn the basics very clearly and gain the confidence to be a developer.
I now have the confidence to crack system design interviews.
Generative AI & Agentic AI Professional Program
- 12 weeks from the first class to the capstone
- 12 modules 167 topics, every one in the syllabus below
- Live in Tamil with recordings and notes in English
- Hands-on every topic tried on your own machine
- LLMs, prompt engineering and LLM APIs (OpenAI, Claude, Gemini, Ollama)
- Production RAG with vector databases, evaluation and fine-tuning
- AI agents with LangGraph, CrewAI, AutoGen and MCP
- A weekly lab, a mini project and a deployed capstone