Hi, I'm Harvey 👋

Engineering lead · Building AI-powered dev tools & products
📍 Changsha, China

---

🧭 What I'm working on

I lead engineering teams building AI-native products and developer tooling. My current focus:

• 🔭 AI coding pipelines — end-to-end workflows that bring real engineering rigor to AI-assisted development: requirement clarification → tech-stack enforcement → TDD → DevOps automation.
• 🛠️ LLM-powered product systems — currently rewriting VOC Insight (a customer-feedback intelligence platform) on NestJS, with a stronger architectural foundation.
• 🧪 Dev workflow augmentation — exploring slash-command toolkits, skill frameworks, and agent patterns that actually hold up in production.
• 📈 Observability for AI systems — designing adaptive crawler & data-collection pipelines instrumented with InfluxDB + Grafana.

⚙️ Tech I work with

const stack = {
  languages:   ['TypeScript', 'Python', 'Java'],
  backend:     ['NestJS', 'Node.js', 'FastAPI'],
  ai:          ['LLM orchestration', 'Agent workflows', 'Prompt engineering'],
  data:        ['MySQL', 'Redis', 'InfluxDB'],
  philosophy:  'Design-first. Pseudocode before code. Ship with rigor.',
};

💡 How I think about AI tooling

The interesting question isn't "can the model write this code?" — it's "what scaffolding turns a model into a reliable engineer?"

I care about the boring parts: spec clarity, test coverage, deployment hygiene, observability. AI assistance only compounds when the surrounding pipeline is disciplined.

🌱 Currently exploring

• Skill / slash-command frameworks for AI coding agents
• TDD-driven AI workflows (tests as the spec the model writes against)
• Adaptive strategy systems for long-running data pipelines

📫 Get in touch

• ✉️ Email: [email protected]
• ✉️ Email: [email protected]

---

🧘 Deep focus mode: ON · ⚡ Vibe coding with discipline