Bringing AI to an exciting untapped market. Stay tuned

I improved the developer experience for all internal engineers, and improved the infra for AI evals.

● Improved the postmortem authoring process for a 200+ person organization. Reduced the average incident-to-publish turnaround time from over a month to a few weeks.

● Simplified mobile device management for Amazon's 1,000,000+ Android devices used in fulfillment centers globally.

● Converted server monolith to microservice architecture. Introduced and taught microservice best practices.

● Responsible for full-stack data AuthN/AuthZ, and automated tools to ensure privacy of sensitive user data in CI.

● Created multi-language autogenerated SDK and documentation for internal and public APIs. This simplified customer product API integration, eased customer onboarding, and reduced cross-service/integration bugs.

● Responsible for database administration (performance monitoring/improvements, security) across the full backend stack, designed and implemented data models across all products.

● Regularly performed security audits of full-stack packages, preventing multiple vulnerabilities from being deployed.

● Introduced a formal design-review process, raising quality of final system architecture, reducing bus factor, and improving code quality.

Python (FastAPI, SQLAlchemy, Celery), NodeJS, Typescript, PostgreSQL, Redis, AWS, Bash/Zsh, Linux, macOS

● Maintained core microservices (Python/Typescript) handling thousands of requests/second delivering 100+gb assets related to autonomous vehicles' various maps. Ensured service uptime and error rates were unaffected during project rollouts via traffic shadowing.

● Automated a rasterization pipeline (in Python, using Luigi framework) which transformed TBs of various inputs (raw lidar, semantic features, etc) to GBs of vectors for consumption by autonomous vehicles, reducing manual interventions from 8+ engineering hours down to 0.

● Designed a map/reduce pipeline sharding framework, reducing the rasterization process from over a day down to six hours.

Python, NodeJS, TypeScript, PostgreSQL, Google Cloud Platform, Kubernetes, Bash/Zsh, Linux, macOS