#qdrant-vector-database

I Build a Healthcare retrieval system using GLiNER2, Neo4j, Qdrant, and QQL to convert doctor–patient conversations into structured medical intelligence. Implemented healthcare entity and relation extraction, constructed a knowledge graph, and enabled hybrid semantic and graph-based retrieval for efficient healthcare data search and analysis.

OpenAI, Qdrant, Cohere Reranker -- grounded RAG pipeline for messy legal documents — verbatim-quote-grounded first-pass memo + typed operator-edit learning loop. FastAPI + React + Docker.

SwarmPIM is a production-grade, containerized PIM system engineered with Next.js 14 and FastAPI. It orchestrates a multi-agent AI swarm utilizing Gemini and Grok with dynamic fallback routing. The architecture features an embedded Qdrant vector cache, high-speed Polars batch ingestion, and analytical DuckDB storage optimized via Apache Arrow.

A hybrid RAG assistant built on LangGraph, FastAPI, and Streamlit which also works with langgraph studio. Integrating parallel map-reduce routing, real-time token streaming via langgraph event, and a 5-step loop guard reflection mechanism for highly accurate, fact-checked responses.