AI Season — AI Agents Bootcamp: Course Material
Runnable lesson code from a live AI agents bootcamp taught in Urdu and English, covering LLM API basics, retrieval-augmented generation, building an agent harness from scratch, multimodal agents, LangChain, agent-to-agent protocol, and training a small language model.
This repository holds runnable code from Cohort 01 of the AI Season AI agents bootcamp, a live online course taught in Urdu and English. Each folder is a self-contained, numbered lesson with its own README and requirements file, written as flat, readable teaching scripts rather than a framework students must learn first.
Sessions cover first LLM API calls and streaming, RAG over documents with chunking, embeddings, hybrid search and a Streamlit chat app, building an agent harness of tools, memory, safety rules, a verifier and tests checkpoint by checkpoint, multimodal agents over vision, audio, video and OCR, LangChain prompt templates and chains with a leads agent, production patterns such as a RAG system that refuses to answer when unsure and a self-correcting coding agent, agent-to-agent delegation and a two-agent debate using the official A2A SDK, building a small language model from scratch and serving it, and a WhatsApp bot built with LangChain.js.
Built with Python and JavaScript, using LangChain, FAISS, Groq, the Gemini API and PyTorch depending on the session.