We are hiring an AI Integration Engineer to integrate MedGemma into our on-prem Hospital Information Management System (ZypoCare by Zyposoft).
You will build/extend an AI microservice (FastAPI preferred) to provide context-aware clinical drafting and a screen-aware AI Copilot across HIMS modules. The integration must be secure (RBAC + audit logs), reliable (streaming chat, retries, timeouts), and support RAG with our internal SOPs/templates.
Responsibilities
Integrate MedGemma for local inference (GPU-ready, CPU fallback)
Build RAG pipeline (embeddings + vector store + citations)
Implement secure APIs (/chat streaming, /summarize, /extract)
Enforce branch + role-based access, PHI-safe logs, full audit trails
Integrate with Next.js web app (screen-aware summaries, copilot panel)
Add observability + performance tuning + documentation/runbooks
Must have
Production LLM integration experience (HuggingFace/vLLM/Ollama/llama.cpp)
FastAPI + Python, Docker, Postgres/Redis
RAG + vector DB experience
Security mindset for sensitive data
Deliverables
Working AI copilot integrated in our HIMS with RBAC, audit logs, RAG citations, streaming chat, and deployable Docker setup.