Projects
Selected work, ongoing and shipped.
-
Clinical RAG pipeline
Production RAG system processing 12M+ medical records per day — OCR, document classification, hybrid retrieval, and grounded answer generation for clinical data access.
-
Internal MCP servers for text-to-SQL
Model Context Protocol servers with auth, scoped tools, streaming, and schema validation — powering multi-step text-to-SQL flows for enterprise analytics.
-
LLM analytics & cost observability
Dashboards and eval tooling for tracking LLM performance, hallucination rate, and cost — cut infrastructure spend by ~45% across deployed GenAI workloads.
-
Genomics pipeline on AWS
Orchestrated BAM/VCF processing on AWS HealthOmics with Airflow, DynamoDB metastore, SQS/SNS, and Lambda triggers. Plus a self-serve file-discovery service for sales and delivery teams.