.ap

PharmaAI

Project details

RAG answers grounded in a curated medication DB; designed for offline/local use with low energy footprint.

  • Sentence-embedding search + context packing
  • Prompt templates to reduce hallucinations
  • Streamlit conversational UI; Dockerized

Results & impact

  • 92% response accuracy · ~2.8s median latency · 3% hallucination

Tech stack

OllamaEmbeddingsRAGStreamlitDocker
Abhimanyu Pandey - ML + Full-Stack