Shubham Shetty

A persistent relational database built from scratch in C++17.

Database internals with a live SQL shell

Faster LLM serving for RAG by extending CacheBlend with semantic-aware KV cache reuse.

Semantic KV cache reuse for faster RAG inference

A free, local-first tool that pings you on your phone when an AI coding agent finishes, errors, or needs an answer.

Two-way notifications for AI coding agents

An NFC-powered travel gallery that turns a fridge magnet tap into a spinning globe flight and a photo gallery.

A globe that flies to your next trip

An AI receipt splitter that turns messy grocery and delivery receipts into a fair settlement.

AI-parsed receipts, split fairly

A privacy-first Chrome extension for highlighting web pages and jotting floating sticky notes.

Floating notes for the whole web

Palo Alto, CA

• Working on query and planning optimization inside the C++ and Python codebase that powers Redshift's query planner.

Redwood City, CA

• Engineering a Premise Awareness feature for a San Mateo law enforcement platform, fusing data across 16 police departments so first responders see hazards, photos, and prior incidents before arriving on scene.

Mumbai, India

• Developed and deployed a GPT-4 powered natural language to SQL interface that let non-technical users query complex data lake schemas, cutting analytics turnaround time by 70%.
• Spearheaded ETL pipelines on AWS using PySpark and Pandas, processing data from 85+ sources and over 20GB daily, saving $150K+ a year by decommissioning legacy workflows.
• Engineered a reusable data ingestion framework supporting multiple sources and formats on AWS Lambda, Glue, and S3, cutting onboarding time for new data sources by 80%.
• Architected the backend using Spring Boot with Aurora PostgreSQL on AWS ECS, building REST APIs for 10,000+ production users and optimizing database queries for performance.
• Engineered a blue-green deployment strategy for ECS services with Terraform and Spinnaker, enabling zero-downtime deploys and rollback for critical production workloads.
• Built a user-facing dashboard with Java Spring Boot and ReactJS on AWS ECS, using AWS ElastiCache for caching to support 1,000+ active users.

Mumbai, India

• Promoted to Software Engineer II in 18 months, in the top 3% of new hires globally, for high impact on critical data lake initiatives.
• Built a data lineage tracking solution for the data lake using AWS Neptune, improving data quality and observability and cutting compliance audit time by 40%.
• Authored a technical primer on using Liquibase for Redshift that drew over 700 unique visitors in 3 months and improved team knowledge.