Applied AI / Enterprise System
Private / NDAViriva Medical RAG
Built an enterprise RAG medical report analysis chatbot using FAISS vector search and open-source LLMs (LLaMA/Mistral), reducing manual report interpretation effort for end users.
Contact for a walkthrough ↗AI Intern
Role
Private / NDA
Access
01 — Problem
Medical professionals and end users faced significant overhead manually interpreting dense, complex medical reports.
02 — Context
Built at Viriva AI Private Limited, where streamlining data extraction and interpretation directly impacts user efficiency and decision-making speed.
03 — System Built
An enterprise Retrieval-Augmented Generation chatbot designed to analyze medical reports, answer queries, and extract key insights safely.
04 — Engineering
The full architecture and dataset are covered under an NDA with Viriva AI — get in touch for a walkthrough under confidence.
05 — Decisions
Chose open-source LLMs over proprietary APIs to keep medical records inside the deployment boundary, and FAISS for local, high-speed vector retrieval instead of a heavy managed vector database.
06 — Result
Reduced manual report interpretation effort for end users.
07 — What I Learned
Deepened understanding of context-window limitations in LLMs and the importance of clean chunking strategies in medical data, where context separation can alter meaning.
08 — Next Iteration
Implement a re-ranking pipeline and confidence scoring to prevent hallucinated medical assertions.