AI & RAG Architecture

🧠 LangGraph Multi-Agent, Model Context Protocol (MCP) & RRF RAG

Architecting multi-agent LLM systems with LangGraph, Model Context Protocol (MCP), Reciprocal Rank Fusion (RRF), and Cohere Rerank 3.

1. Hybrid BM25 + Vector Search with Reciprocal Rank Fusion (RRF)

# Reciprocal Rank Fusion (RRF) Math Formulation
RRF_Score(d) = sum( 1 / (60 + rank_dense(d)) + 1 / (60 + rank_sparse(d)) )

2. Model Context Protocol (MCP) Integration

MCP acts as a universal context and tool execution layer, enabling agents defined in LangGraph to interact with external databases and APIs safely.

3. Interactive AI Labs & Real Guides:

• ⚡ Interactive Neural Network & Softmax Simulator Suite
• ⚡ CNN 3x3 Spatial Convolution Filter Kernel Studio
• ⚡ LLM Tokenizer & Self-Attention Heatmap