qwen-local: Running an OpenAI-Compatible Model Service on Apple Silicon
qwen-local is a thin FastAPI service around MLX models for chat, embeddings, Qwen3 text-to-speech, and Whisper speech-to-text on a 16 GB Mac.
Tailgate: A Private AI Gateway for Local and Remote Models
Tailgate is my private OpenAI-compatible gateway for routing AI clients across local and hosted providers.
PocketBabel: Browser Translation Without a Backend
PocketBabel is a small English-Chinese translation app that runs transformer models directly in the browser.
Minifish Lab Projects
A living map of the small systems, tools, experiments, and prototypes I keep under Minifish Lab.
Running OpenAI Symphony as a Solo Developer Across Two Repos
A hands-on report on running OpenAI Symphony with Linear, GitHub, and Codex across two personal repositories.
From AI Conversations to Published Blog: The MCP-Powered Publishing Revolution
A practical publishing workflow that turns AI conversations into Hugo posts through MCP, GitHub Actions, and a structured blog repository.
Brikka on Induction: An Engineering View of a Triggered Extraction System
A systems view of brewing with a Brikka on induction, treating extraction as a triggered thermal and pressure process.
OceanBase Internals: Transactions, Replay, SQL Engine, and Unit Placement
OceanBase targets high availability and scalability in a shared-nothing cluster. The core engineering challenge is to make four critical subsystems work together with predictable…
Introduction to WebVM
WebVM is a virtual machine (VM) that executes entirely within a web browser. It’s an innovative project that brings the power of a Linux environment straight to your browser,…
Notes on Modifying a Brompton Clone Bike
Notes on modifying an Aceoffix Brompton clone bike, including accessories, tradeoffs, cost, and practical buying tips.