I'm a master's student working across infrastructure, full-stack software, and computer vision. I like designing systems from the interface down to the API, data, workers, and deployment. My current research focuses on Sim2Real vision for sports, especially volleyball and basketball. If a tool I want doesn't exist, I often build it.
- Infrastructure & system design: containerized services, networking, storage, CI/CD, Terraform, and distributed task processing—including parallel GPU workers on Kubernetes.
- Frontend & backend: web and desktop interfaces, APIs, databases, real-time communication, and video workflows.
- Computer vision & simulation: synthetic data, detection and tracking, pose, camera geometry, multi-camera video, and sports analysis.
- Tools for my own needs: small apps, CLIs, and developer tools that make a specific workflow easier.
I've come across these in projects and experiments. My experience varies by tool, and I'm still learning many of them.
| Area | Tools |
|---|---|
| Languages |             |
| Web frontend |     |
| Full-stack & SSR |    |
| Backend & APIs |         |
| Data & storage |       |
| Infrastructure & delivery |          |
| Desktop & mobile |     |
| Vision & ML |    |
| Simulation & game engines |   |
| Media & real-time |      |
Ranked using code size and repository count; notebooks and build-file formats are excluded. This reflects public code, not proficiency.
GitHub activity
Vision & simulation
- Volleyball Analysis Engine — detection, tracking, pose, and analysis in a Python / PyTorch / OpenCV / Docker worker.
- NeuralCourt Basketball — a Unity / C# basketball simulation with GPU-batched character motion and multi-camera monitoring.
Full-stack applications
- Volleyball Monitoring AI — live annotation and video workflows across Nuxt / Vue / TypeScript, Fastify / GraphQL Yoga / Prisma / PostgreSQL, and Redis / MinIO / WebSocket / FFmpeg / OvenMediaEngine / Traefik.
- PatentsHub — a desktop application using Electron / Nuxt / Vue / tRPC / Prisma.
Tools I built because I needed them
- TartUI — a SwiftUI interface for managing Tart images and virtual machines on macOS.
- k8s-mig-manager — a Bun / TypeScript / Ink CLI for NVIDIA MIG profiles on Kubernetes nodes.
- Snip — a SwiftUI / yt-dlp / FFmpeg app for downloading and clipping video.
- Cloudflare Catch-all Mailbox — a searchable inbox API built with Workers / D1 / R2.
- Nuxt Imperative Modals — a Nuxt / Vue / TypeScript module for opening typed modals from code.
More experiments and projects are on my repositories page.