At Datum Bio, I architected and scaled a pharmacogenomics knowledge graph exceeding 500M triples, modeling CPIC, FDA, and DPWG guidelines for clinical decision support. I also designed SPARQL pipelines to unify pharmacogenomic and clinical data.
I build GraphRAG and vector-embedding retrieval pipelines that ground LLM applications in structured knowledge. At Datum Bio, I delivered full-stack features across three shipped products, including pgx.datum.bio and beta.mypgx.org.
As an independent consultant, I design Neo4j graph models and Cypher ingestion workflows for large datasets, and migrated a client's relational knowledge base to a graph-native architecture. I also migrated an on-premise MySQL database to Azure-hosted MySQL with zero data loss.
Earlier, I maintained production databases at Cotiviti Nepal Pvt. Ltd. and built streaming data pipelines. At Wiseyak Inc., I built an ontology-driven clinical decision support system that went live at Dhulikhel Hospital.
