AI Engineering & Agent Development
AI engineering for the web, hands-on - local-first LLM coding agents, Claude Code plugins, MCP servers, browser-native ML, and the occasional experiment in breaking my own models. This page tracks what I'm shipping into AI-enabled web development.
Current Focus
- -Local-first coding agents (Ollama, Gemma, ReAct loops)
- -Claude Code plugins & authored skills
- -Model Context Protocol (MCP) servers for real dev workflows
- -Adversarial examples & model robustness in NLP
- -Browser-native inference with ONNX Runtime
Tools & Experiments
nip-session.sh
$ nip "refactor getUser to async/await and add a cache"
> reading src/api/users.js ... [ok]
> plan: promisify callback, wrap with async, add LRU layer
> applying patch (3 hunks) ... [ok]
> running tests ... 12 passed, 0 failed.
> agent status: idle.