LM: A Programming Language Where LLMs Write Fewer Bugs
Can language design eliminate the bugs LLMs make most often? We tested LM against TypeScript and Python in a blind benchmark — and found that compiler constraints matter more for weak models.
Writing about software engineering, quantitative finance, and building things.
Recent Posts
I Built a WeChat Channel Plugin for Claude Code
claude-code-weixin brings Claude Code to WeChat — chat with your AI coding assistant directly from your phone, with support for text, images, files, voice, and video.
今日新闻简报 | 2026年3月14日
战火与博弈主导全球议程：美伊冲突白热化、全球股市承压、两会定调AI十万亿、美国301调查围堵中国、乌美俄和谈再度搁浅。
Building a Multi-Factor Quant Trading System in Python
How I built a quantitative trading framework that combines six alpha factors with mean-variance portfolio optimization and automated execution via Alpaca.