corrective-rag

How to build a simplified Corrective RAG assistant with Amazon Bedrock using LLMs, Embeddings model, Knowledge Bases for Amazon Bedrock, and Agents for Amazon Bedrock.

Production-grade RAG system with hybrid retrieval (Qdrant + Elasticsearch + Neo4j), Corrective RAG via LangGraph, feedback-driven reward model, and RAGAS evaluation dashboard

Agentic RAG system with LangGraph, hybrid BM25+FAISS retrieval, cross-encoder reranking, Corrective RAG, FastAPI, RAGAs evaluation, and Docker deployment

ATLAS — industrial Agentic RAG platform (LangGraph): routing, query decomposition, Corrective RAG, Self-RAG reflection. FastAPI + SSE, pluggable vector stores, auth, rate limiting, tracing, Docker, CI. Academic webinar kit on archive/academic-labs branch.

RAG enhances LLMs by retrieving relevant external knowledge before generating responses, improving accuracy and reducing hallucinations.

Self-correcting agentic RAG with LangGraph — combines Adaptive, Corrective, and Self-RAG. Routes queries, grades retrieved documents, and checks answers for hallucinations. Runs fully local on Ollama.

A complete, structured RAG bootcamp covering every layer of modern AI pipelines, data ingestion, vector search, agentic architectures, memory, guardrails, and real-world evaluation. Built with LangChain, LangGraph, and Python.

Self-updating research assistant using RAPTOR, corrective RAG, and arXiv retrieval.

Corrective-RAG flood and disaster relief assistant for Assam, grounded in 27 official NDMA/ASDMA/CWC/IMD documents, with a live situational map computing real-time flood risk from rainfall and river discharge data on satellite imagery. Falls back to government-only web search when retrieval is unreliable.

Corrective RAG with LangGraph: evaluates retrieval quality, routes to web search when needed, refines context, and generates grounded answers.

CRAG with MinerU

An engineering-oriented Agentic RAG system built with FastAPI, LangGraph and Qdrant, featuring multi-user document isolation, hybrid retrieval, reranking, corrective retrieval, document-version-aware conversations and streaming Web UI.

Agentic RAG system with five workflow modes from Naive to Deep, routing each question to the cheapest path that answers it well, fusing dense and BM25 retrieval through Reciprocal Rank Fusion with CrossEncoder reranking, verifying groundedness and citations, and gating risky answers behind a human-in-the-loop approval queue.

adaptive rag, corrective rag and agentic rag examples using langgraph

A research notebook that answers only from the sources you give it. Add a PDF, a YouTube link or a web page, then ask: every sentence cites the exact page, timestamp or character range behind it. Hybrid retrieval, reranking and a corrective loop that refuses rather than guesses.

It is a enhanced version of Past Portals with Multi-Modal Input system , C-RAG , Feedback Loop, and Voice-First Conversational AI bot

Successfully developed a Healthcare AI Clinical Decision Support System, leveraging LangGraph, GPT-4o-mini, and PubMed to deliver real-time patient risk stratification, evidence-based treatment recommendations, and personalized clinical road maps with integrated drug safety validations.

Automated Agentic GitHub PR review bot — GPT-5 agentic system with 3 tools: Corrective RAG (project context), MCP web search (live docs), and ruff linter. Redis-Celery task queue. Structured review comments posted automatically on every PR.

A RAG agent that checks its own answers before trusting them and searches the web instead of guessing when its knowledge base falls short.

Runnable LangGraph, CRAG, and GraphRAG lab tested against a production hybrid-RAG baseline.