All projects
January 2025
TrustMed-AI — Production Multi-Agent Medical RAG System
GenAIMulti-Agent AIRAGHealthcare AIProduction ML
- Medical Q&A system that answers clinical questions using only verified medical sources, with a 3D avatar that speaks answers aloud for accessible patient communication.
- Multi-agent RAG framework with ThreadPoolExecutor parallelization lifts retrieval precision 40% over single-agent baselines while cutting response latency 12s → 4s.
- FAISS cross-encoder reranking on a privacy-preserving pipeline reduces hallucinations 65%, achieving 91% accuracy on a MedQA-style benchmark across 5 medical domains.
- 100-prompt evaluation suite logged to Snowflake and visualized in Tableau tracks precision and faithfulness; Next.js frontend with Three.js avatar delivers multimodal text-to-speech responses.
PythonLangChainFAISSGPT-4o/OpenAI APIFastAPIChromaDBThree.jsDockerAWSNext.jsSnowflakeTableau