Yaqin Hei

I’m an AI engineer focused on shipping large language models and multi-agent systems into production. At Apple I built conversational AI serving millions of users, specialising in continual learning and domain adaptation for LLMs, and led a team building multi-agent systems that automated 70% of IT tasks for Fortune 500 companies.

The gap between AI research and production is enormous. I write two kinds of essays: methodology, distilled from projects that shipped, and postmortems, with the logs to prove them. Most are available in English and Chinese. I also built AutoContent, which takes a piece from topic discovery to WeChat publication in 30 minutes.

What I write about

  • LLM fine-tuning and continual learning in production
  • Multi-agent system architecture and orchestration
  • Deploying AI systems at scale
  • Bridging research and production engineering

Agent Audit · Free

Send me your agent’s worst conversation. I’ll tell you why it broke.

I built and audited production agents at Apple and beyond. Paste the transcript, and you’ll get back a specific diagnosis — architecture, guardrails, or eval — and how to fix it. Free, no strings.

  1. 1

    Send the transcript. The uglier the failure, the better the teardown.

  2. 2

    Get the diagnosis. Where it broke: intent, tools, critic, or knowledge.

  3. 3

    Get the fix. Concrete next steps, mapped to the essays.

Elsewhere

微信公众号 · 京墨AI研习社 @HeiLabAI — 中文长文首发
视频号 · Yaqin.AI Video walkthroughs of selected essays
其他公众号 商树实验室 @BusinessTreeLab · 玛塔塔成长志 @MatataGrowth
All original content is licensed under CC BY-NC-SA 4.0.