AI Infrastructure, Model Training Orchestration Platform

• Implemented POC design for a resource manager, which assigns compute resources to distributed big data compute sessions within Meta AI/Spark infrastructure. Scale of 215M jobs and 15k users daily

• Optimized MySQL database, Thrift network, Java software performance to bring down e2e latency while supporting 10x vertical scale (100k to 1M+ concurrent sessions)

• Gathered metrics and bottleneck data to assess viability and strategic next steps for prototype

• Designed and implemented "save as draft" feature for core product, used by 30+ customer organizations

• Worked directly with technical customers to build SIEM log publishing feature for security-relevant events impacting all users

• Collaborated with PMs, engineers, and designers in design discussions, feature planning, company-wide retrospectives

• Worked with React, Typescript, and Tailwind to collaboratively implement various public-facing static sites. Implemented animations manually and with various animation libraries.

• Built interactive, animated hacker applications tool used to submit 8000+ applications

• Built RSVP tool and integrated with our backend to support 1000+ hackers in real time

• Participated in design reviews and discussions, as well as process and project retrospectives

• Designed software and process to support product boosting, which complements ML-based product ordering along Uber's core ride request flow. Impacts ~525 million ride requests per month worldwide

• Restructured boosting configuration to allow for better governance, monitoring, and data aggregation. Oversaw migration of ~100 currently boosted products between configurations.

• Iteratively designed and implemented UI/UX flow to reshape internal processes for requesting product boosts

• Consulted with operations, product, backend engineers and ML engineers to scope, design, and deliver tooling over the course of the internship