Stanford AI Lab–inspired rigor, shipped to production. This portfolio highlights RAG LLMOps for travel search and GCP-native BigQuery data marts for analytics—built with tight engineering loops, objective evals, and clear paths to value.
Retriever-augmented generation for travel planning: ingestion → vector index → semantic search + re-rank → grounded answers with citations and guardrails.
Ultra-short walkthrough showing retrieval, re-ranking, and answer generation loop with telemetry for rapid iteration.
Governed ingestion, Bronze→Silver→Gold modeling, Python ETL (batch + incremental), validation contracts, schema evolution, Composer orchestration, Terraform IaC.