I'm Chris Wu — CS @ USC '27, CEO at anvol.inc, building Lovable for hardware. Previously cofounder & CTO at Moorecubes, ex-PM at articuler.ai. I build AI-native products end-to-end: hardware ↔ backend ↔ UX ↔ growth. Self-described vibe coder.
📫 wuhaipen@usc.edu · 🌐 chris-wu.com · 💼 LinkedIn · ✍️ Medium
Open to SWE / PM / GTM-engineer roles — Summer & Fall 2026.
Two robotic products shipped: an autism-intervention robot and a psychological-counseling robot. Built the real-time audio pipeline (ASR → LLM → TTS over WebSockets), the SpringBoot backend, and the WeChat mini-app.
- Stack: Node.js · FastAPI · PostgreSQL · Prisma · WebSockets · Docker · Kubernetes · GitHub Actions
- Highlights: >700K RMB sales · 300K RMB grand prize at Suzhou pitch · YC China alumni (Miraculus Plus) · 200-person UX research study
- Public demo: ChrisWu132/volcEngineDemo — live Volcengine ASR→LLM→TTS slice
- Full video: chris-wu.com/projects/moorecubes
Claude Buddy — CC Desk Pet · w/ @leezisheng · 2026-04 → ongoing
An ESP32 desktop pet that animates Claude Code tool calls in real time over BLE. Press a physical button to approve risky actions — hardware as a trust boundary for the agent.
- Repo: ChrisWu132/MicroPython_Claude_Assistant
- Team: co-built with @leezisheng, who led the device firmware — WS2812 light engine (hardware-timer ISR), Doubao-TTS voice queue, 8 preset animated characters, SD-card logging, and the
flash_device.py/gen_voice_assets.pytooling. I built the Claude Code hook daemon, BLE bridge, and plugin packaging. - Stack: ESP32 · MicroPython · LVGL · BLE (Bleak) · Python asyncio
- Highlights: real-time tool-call streaming · risk-aware physical approval gate · 49 tests · <4 min onboarding
GenAI networking platform for students. Authored 3 PRDs across a cross-functional team; ran LLM evals end-to-end.
- Stack: GenAI · LLM evaluation · cross-functional PM
- Highlights: surfaced a $2M market gap · defined 10 LLM eval standards · +50% feature adoption · +30% output quality
K-12 AI-literacy platform: 6 progressive labs teaching how LLMs work, plus a visual AI-workflow builder.
- Repo: ChrisWu132/labLM · Live: lab-lm.vercel.app
- Stack: Next.js 15 · React Flow · Supabase · TypeScript · Tailwind
- Highlights: Vercel Hackathon 2nd place (USC Vibeathon)
Other public work: axtree-devtools (a11y tree visualizer for GUI agents) · ThroughmyAI (voice-controlled IG automation) · telos (AI learning platform). The portfolio site (chris-wu.com) and the Moorecubes core codebase are private — full breakdowns + live demos live on the site.
| Layer | Tools |
|---|---|
| Frontend & UI | Next.js 15 · React · TypeScript · Tailwind · Framer Motion · React Flow · shadcn/ui |
| Backend & AI | Python · FastAPI · Node.js · LangGraph · MCP · Claude / Gemini APIs |
| Hardware & Edge | ESP32 · MicroPython · LVGL · BLE |
| Infra & Data | PostgreSQL · Supabase · Docker · Kubernetes · GitHub Actions · MongoDB |
| Browser & Automation | Playwright · WebSocket · Puppeteer |
I write about AI, browser agents, and prompt engineering on Medium.
- How Can AI Use Your Browser? A Deep Dive into Automation Frameworks
- Think Agents Are Too Hard? Master Prompt Engineering in a Weekend (Beginner-Friendly)
⚡ USC ping pong varsity captain · cooking obsessive · ex–Softbank China VC analyst (10+ robotics / AI pitches evaluated).
