TheKey
Jan 2024 - Present
Generative AI Engineer
- Built LLM-powered features for internal healthcare tools, allowing users to query patient-related data and workflows in a more natural way instead of navigating multiple systems.
- Designed and iterated on RAG pipelines combining structured data and unstructured documents, improving retrieval quality and reducing manual lookup.
- Implemented backend services in Python (FastAPI) to support real-time AI responses, including streaming outputs and session-aware interactions.
... More
- Integrated embedding-based retrieval using vector search, tuning chunking strategies and retrieval logic to improve consistency across similar queries.
- Worked closely with product to refine prompt design, response formats, and guardrails, improving reliability and usability in real-world workflows.
- Implemented authentication, session management, and access control, ensuring secure handling of sensitive healthcare data.
- Identified bottlenecks in retrieval and generation pipelines, improving latency through async processing and lightweight caching.
- Structured the system to support extensible AI use cases (chat, summarization, workflow automation) without major refactoring.