AI systems, full-stack products, and developer tools
EECS @ UC Berkeley | Founding Software Engineer @ 1Sphere AI
I build small, practical tools around AI workflows, browser productivity, developer analytics, and student infrastructure. Follow along if you want to see shipped products, public build notes, and OSS-friendly projects instead of only private coursework.
Currently focused on:
- AI writing and browser automation tools
- LLM evals, prompt regression testing, and agent workflows
- GitHub/profile analytics and developer-facing dashboards
- Student productivity systems for deadlines, course planning, and collaboration
| Project | What it does | Stack |
|---|---|---|
| snapwrite-ai | Chrome extension for improving, replying, rewriting, summarizing, fixing grammar, and translating selected text on any website. | JavaScript, Manifest V3, Node.js, Playwright |
| arena | LLM regression-testing loop for prompts: mine traces, run evals, judge outputs, optimize variants, and gate PRs. | Python, Typer, Next.js |
| devmetrics | No-auth GitHub analytics dashboard for repo, language, and activity insights. | JavaScript, GitHub API |
| deadline-aggregator | Student deadline dashboard that syncs Canvas, Calendar, Gmail, Gradescope, and Ed Discussion. | TypeScript, Next.js |
| TransferAble | Course planning and transfer roadmap platform for California community college students. | TypeScript, Next.js, Prisma |
| ElamiteDatasetLab | Reproducible Word2Vec and network-analysis lab for ancient Elamite text corpora. | Python, NLP, data science |
- snapwrite-ai issues: Chrome extension UX, provider support, privacy/security hardening, docs, and tests.
- arena issues: LLM eval examples, standalone usability, docs, and CI gate improvements.
- devmetrics issues: GitHub API edge cases, dashboard polish, and shareable profile reports.
- Building LLM agent systems at 1Sphere AI as Founding Software Engineer
- Leading front-end education at Berkeley Open Project @ UC Berkeley
- Polishing a smaller set of useful open-source projects instead of maintaining noisy demo repos
- Following a daily loop: one meaningful upstream OSS contribution, one repo polish commit, one visible product increment
Open to software engineering, AI engineering, and open-source collaboration.