Venu Dave

About

Amazon internship experience building and shipping an end-to-end NL-to-SQL system: ingested/normalized metadata across 60+ internal tables, added rigorous multi-layer validation for LLM-generated SQL, and served it via a FastAPI backend for engineers—driving 90%+ faster dataset discovery and ~70% lower effort to access data. Also built an early-stage RAG-based healthcare assistant, iterating on chunking, embeddings, and retrieval to improve answer quality post-launch.

Key Strengths

• Owned end-to-end production data/ML pipeline (ingestion → validation → transformation → serving)
• Built NL-to-SQL pipeline over 60+ internal tables with schema-context ingestion
• Multi-layer validation for LLM-generated SQL (schema checks, syntax validation, controlled execution, fallback regeneration)
• Delivered measurable impact: 90%+ reduction in dataset discovery time and ~70% less effort to access data
• Designed resilient ingestion with idempotency, retries, and safe reprocessing/backfills
• Shipped internal FastAPI service exposing NL-to-SQL with performance and reliability optimizations
• Iteratively built and improved an early-stage RAG healthcare assistant (chunking/embeddings/retrieval tuning)

Reference Highlights

• Strong end-to-end ownership of data pipelines
• Excellent reliability and data-quality outcomes
• Significantly reduced query failures
• Improved data accessibility
• Strong Python and SQL in real production work
• Ensures correctness via validation and edge-case handling
• Effective debugging to maintain reliable outputs
• Improved system performance by reducing unnecessary joins/context
• Built stable pipelines that reduced manual intervention
• Strong logging and monitoring practices
• Designed reliable idempotent pipelines enabling safe retries/backfills
• Prevents downstream issues through validation and schema handling
• Highly effective generalist in ambiguous environments
• Scopes problems well and makes practical tradeoffs
• Ships end-to-end with minimal direction
• Consistently follows through to improve systems

Experience

Education

Skills

Languages