I'm a software engineer focused on edge computing, IoT systems, cloud-native platforms, and AI-assisted development workflows.
I like building systems that connect the messy physical world with reliable software abstractions: devices, agents, Kubernetes, automation, and developer tools.
Most of my work sits around one question:
How can we make complex device integration and software delivery feel simpler, safer, and more repeatable?
- 🧩 Building cloud-native edge systems for IoT / physical AI scenarios.
- ⚙️ Working with Go, Kubernetes, Docker, gRPC, React, TypeScript, and cloud infrastructure.
- 🤖 Exploring agent-native development workflows with Claude Code, Codex, ChatGPT, and custom coordination tools.
- 🛠️ Interested in developer experience, reliable automation, device onboarding, and AI-assisted code generation.
- 🌍 Open-source contributor and maintainer around edge / IoT infrastructure.
-
Agent-native engineering workflows
Turning messy product ideas into specs, tasks, code, reviews, tests, and release notes with AI agents. -
Edge AI / Physical AI runtime
Making devices easier to detect, configure, test, and integrate into real software systems. -
Cloud-to-edge platforms
Reliable orchestration, deployment, storage, observability, and operations across constrained environments. -
Developer tooling
Building small tools that reduce repetitive engineering work and make code review / debugging loops faster.
| Name | What for |
|---|---|
| Claude Code | Main coding agent for deep implementation, refactoring, bug fixing, codebase exploration, and multi-file changes. |
| Codex | Parallel coding agent for delegated tasks, PR-style changes, code review, and isolated experiments in separate worktrees/cloud sandboxes. |
| ChatGPT | Multimodal thinking partner for planning, writing, research, architecture discussion, visual analysis, image generation, and clear document drafting. |
| Frogo | Taskboard for tracking agent tasks, working status, progress, blockers, and handoffs. |
| Infinite Loop | Iteration engine for “keep trying until it passes”: run → test → inspect error → fix → repeat. Best for debugging loops and verification workflows. |
| Hermes | Personal coordinator, dispatcher, and GitHub bot that routes tasks to the right agent, keeps context, reviews PRs/issues online, comments on GitHub, and turns vague goals into executable tasks. |
I care about:
- Clean interfaces over clever abstractions.
- Reliable systems over impressive demos.
- Local-first and edge-capable workflows when possible.
- Documentation that actually helps agents and humans work better.
- Boring infrastructure that fails gracefully.
- AI tools that improve engineering judgment instead of replacing it.
Build the system, then make it explainable enough for both humans and agents to operate.
Thanks for visiting! 🚀