GenAI-Playground ✨

A comprehensive, hands-on repository of AI agents, RAG pipelines, LangChain/LCEL abstractions, prompt engineering strategies, and full-stack Generative AI applications.

---

🔋 Contents

This repository serves as an active playground for exploring and implementing modern Generative AI engineering concepts across both Python and JavaScript/TypeScript ecosystems.

• 🤖 AI Agents & Workflows: Standard LLM calls, function/tool calling, ReAct loop, Planner-Executor pattern, Research agent, and Reflection agent loops.
• 🔍 Retrieval-Augmented Generation (RAG): Ingestion pipelines, text splitters (length, structured, code-aware), vector stores (Chroma, FAISS, in-memory), and advanced retrievers (MMR, Multi-Query, Contextual Compression, Wikipedia).
• 🦜 LangChain & LCEL: Building blocks including LLMs, ChatModels, Embeddings, PromptTemplates, Output Parsers, Chains, and LCEL Runnables (RunnableSequence, RunnableParallel, RunnablePassthrough, RunnableLambda, RunnableBranch).
• 📄 Structured Outputs: Schema-enforced JSON extraction using Pydantic, TypedDict, Zod schemas, and LangChain structured outputs.
• 🎯 Prompt Engineering: Zero-shot, Few-shot, Persona, and Chain-of-Thought (CoT) techniques.
• 🚀 Full-Stack Applications: Modern web apps built with Express/TypeScript backends and Next.js UI clients for structured outputs, search agents, and Light RAG indexing.
• 📚 In-Depth Notes: Conceptual documentation and architecture diagrams covering core LLM theory, prompt design, agent patterns, LangChain mechanics, and RAG pipelines.

---

🐍 Python Development Guide

1. Create a virtual environment

python -m venv venv

2. Activate the virtual environment

Linux / macOS

source venv/bin/activate

Windows

venv\Scripts\activate

3. Install project dependencies

pip install -r requirements.txt

4. Run the development server of the FastAPI application

uvicorn main:app --reload

5. Update dependencies (when required)

pip freeze > requirements.txt

---

Learning by building, experimenting, and iterating. 🌱