Month 1 — LLM Foundations, Transformers & Practical GenAIWeeks 1–3
Transformer architecture (attention, self-attention, encoder/decoder), tokens, embeddings & positional encoding, context windows & inference, proprietary vs open-weight models, prompt engineering (zero/few-shot, structured), HuggingFace Transformers, function calling & structured outputs, LLM pipelines, prompt chaining, guardrails, cost optimisation. Project: multi-step LLM workflow.
Month 2 — RAG, Fine-Tuning & Production PipelinesWeeks 4–6
RAG architecture, chunking strategies, embeddings & vector stores (FAISS, Chroma, LlamaIndex, LangChain), semantic vs keyword search, reranking, metadata filtering, Graph & multimodal RAG, fine-tuning (SFT, LoRA, QLoRA, PEFT), distillation & quantisation. Projects: cited QA RAG system; fine-tune a small LLM.
Month 3 — Agentic AI, Automation & Tool-Using SystemsWeeks 7–10
Agents vs workflows, planning / routing / memory, tool use, ReAct & reflection, single vs multi-agent (LangGraph, CrewAI, AutoGen), human-in-the-loop, Model Context Protocol (MCP), REST APIs, ETL/ELT pipelines, scheduling & orchestration, deployment, containerisation, monitoring. Projects: controlled tool-using agent; AI automation microservice.