• Founding engineer, shipped end-to-end features supporting real-world club operations using Next.js, TypeScript, and Supabase, working in rapid weekly release cycles with the founding team.

• Centralized Stripe access in a dedicated payments service and added idempotency guards, eliminating double-subscription edge cases and stabilizing billing.

• Built membership-tier dues (quarterly/semi-annual), member application review + withdrawal, and a reusable email-template UI; refactored email pipeline for cleaner responsibilities.

• Integrated PostHog and instrumented key flows to unblock product decisions; partnered with design to launch venue/discovery pages across 7+ hubs (Commons, Nook, DecantSF, etc.).

• Built GovGnome, an AI platform tracking San Francisco legislation in real-time to help businesses

navigate regulatory processes, reduce compliance costs, and increase civic transparency.

• Architected and built a functional prototype for San Francisco government transparency that could

scrape and process 100+ legislative documents daily from the San Francisco Board of Supervisors with 98% accuracy using LLMs, React (Next.js), Typescript, and Supabase (PostgreSQL).

• Performed 40+ user interviews following lean startup methodology, using qualitative analysis to identify key user pain points and translate them into technical requirements for the product roadmap.

• Discovered a $400M TAM in SF for B2B regulatory compliance & permit processing solutions.

• Generated a validated business case for permit processing optimization that demonstrated potential 40% reduction in approval time for local developers, saving them an estimated $25K per project and pivoting away from general transparency tools with limited market traction.

• SWE#5 at the company

• Co-developed a safety-critical system for high-power wireless transmitters that passed UL certification for Toyota's factory deployment, implementing a hybrid computer vision/RF detection system with 99.8% accuracy for human detection, automatically lowering power to ensure FCC compliance and prevent harmful radiation exposure.

• Achieved 96% precision in differentiating between human and non-human obstructions by designing state machine architecture that processed RF signal data in <15ms, applying statistical analysis.

• Engineered TheOne, a distributed simulation framework that reduced QA testing time by 65%, scaling to 40 transmitters and 100 receivers by implementing message queuing with RabbitMQ and

containerized test environments with Docker.

• Mitigated regulatory risk and potential $1M+ in FCC penalties by developing comprehensive safety

validation protocols with 15+ standardized test scenarios across various distances and beam offsets,

supporting successful product deployment in a highly regulated manufacturing environment.