RAG Experiments
Retrieval-augmented generation experiments for grounded, accurate answers over real documents.
Comparing chunking strategies, embedding models, hybrid search and re-ranking to find what actually improves answer quality on real-world documents — measured, not guessed.
- Experiments
- 4
- Open Source
- 4
- Technologies
- 12
- More to Explore
- ∞
Featured RAG Projects
A few highlighted experiments and prototypes.
- AI AgentsFeatured
AegiSense AI Assistant
An AI-powered assistant for identity and security operations. Uses RAG, tool calling, and guardrails to answer security questions and automate workflows.
- LangGraph
- FastAPI
- PostgreSQL
- Docker
- RAGFeatured
Document Intelligence Lab
Experiments with different RAG pipelines, chunking strategies, and vector databases to find the best setup for real-world use cases.
- LlamaIndex
- Qdrant
- OpenAI
- PDF/Docs
- AI Agents
Customer Support Agent
An AI agent that answers customer queries using company documentation.
- RAG
- FastAPI
- Next.js
How My RAG Pipelines Work
A typical architecture used in these experiments.
- Documents(PDF, docs, web)
- Chunking(Semantic splits)
- Embeddings(Vector store)
- Retrieval(Hybrid + rerank)
- LLM(Grounded answer)
- Citations(Source links)
Evaluate and tune
All RAG Experiments
Explore all my RAG experiments, from simple prototypes to advanced systems.
4 experiments
An AI-powered assistant for identity and security operations. Uses RAG, tool calling, and guardrails to answer security questions and automate workflows.
- LangGraph
- FastAPI
- PostgreSQL
Experiments with different RAG pipelines, chunking strategies, and vector databases to find the best setup for real-world use cases.
- LlamaIndex
- Qdrant
- OpenAI
An AI agent that answers customer queries using company documentation.
- RAG
- FastAPI
- Next.js
Benchmarking dense, sparse and hybrid retrieval with re-ranking on real enterprise documents.
- Python
- PostgreSQL
- Qdrant
Experiment · Learn · Build · Share
Let’s Build the Next Generation of RAG Experiments
Check out the code, try the demos, or get in touch to collaborate on exciting ideas.