I build for the web, the terminal, and the chain β mostly in TypeScript and Go.
- π§βπ» Six years of shipping product frontends (React / Next.js) and backends (Node, Go, Java)
- βοΈ Built Web3 features with Fireblocks WaaS, Ethers.js, and smart-contract integrations
- π€ All-in on LLM-assisted engineering β building my own coding agent to understand the loop end to end
nexteditor.dev Β· source Β· TypeScript
A browser-based recorder and replay engine for real, multi-file coding lessons. It captures the whole workspace β editor changes, live preview, terminal, HTTP calls, slides, captions, audio, even the instructor camera β and replays everything from a single timeline. Recordings stream as .ne files that start playing before they finish downloading, and Yjs-powered live rooms with voice chat make lessons collaborative. Built with Monaco, WebContainers, rrweb, and Cloudflare Workers.
π¬ Its Studio is something that doesn't exist anywhere else yet: an AI lesson-production pipeline inside the editor. A director compiles authored scripts into deterministic performances β typing the code in the real editor, drag-highlighting exactly what the narration explains, running it live β with in-browser TTS narration linted by an editorial critic. Every render must pass mechanical QA and a two-render repeatability check before a human reviews and publishes it.
github.com/channyeintun/nami Β· Go
An agentic coding CLI powered by LLMs β think, plan, and execute code changes without leaving the terminal. A Go engine drives the agent loop, tool execution, and permission gating; a custom TUI (Silvery) streams the session; and plans, task lists, and walkthroughs persist as first-class, reviewable artifacts. Works with Anthropic, OpenAI, Google, DeepSeek, Groq, Mistral, Ollama, and GitHub Copilot.
Frontend
Backend
Delivery
Web3 Β· AI
If Next Editor or Nami looks useful to you, a β goes a long way.




