I'm Dario Zhang, an AI platform lead in Beijing with 8 years in AI. I started in NLP and causal inference.
- Built a low-code agent platform at BlueFocus: 2,000+ agents in production, tools behind an MCP gateway, layered memory for multi-agent context.
- Led KOX, a multi-agent video generation system on Strands Agents and AWS Bedrock AgentCore.
- Now lead an AI production platform: multi-model image and video orchestration, model routing and failover. It went live in production within 3 months.
- My code is written by a team of AI agents. I design the system, split the work and review what comes back.
Side projects: agent-self-evolution · biomorphic-memory · nous · claude-guard
Education: B.S. in Mathematics, Tongji University
Latest Blog Posts (Note & Push)
| Title | Describe | Date | Link |
|---|---|---|---|
| 解决Ubuntu 22.04上APT存储库警告和下载速度问题 | Note & Push | Aug 11, 2023 | https://blog.dariolink.vercel.app/解决ubuntu-2204上apt存储库警告和下载速度问题 |
| 在WSL的vscode上使用GPG签名提交更改 | 因为GPG在WSL上不弹出图形窗口的缘故，需要先在命令行调出密码输入的命令行窗口，然后缓存秘钥，之后才能用vscode的git提交。 | Jun 9, 2023 | https://blog.dariolink.vercel.app/在wsl的vscode上使用gpg签名提交更改 |
📊 This Week I Spent My Time On
💬 Programming Languages:
Other 36 hrs 42 mins ████████████░░░░░░░░░░░░░ 47.98 %
Markdown 20 hrs 57 mins ███████░░░░░░░░░░░░░░░░░░ 27.39 %
Diff 9 hrs 14 mins ███░░░░░░░░░░░░░░░░░░░░░░ 12.07 %
Python 4 hrs 17 mins █░░░░░░░░░░░░░░░░░░░░░░░░ 05.61 %
Text 1 hr 21 mins ░░░░░░░░░░░░░░░░░░░░░░░░░ 01.78 %
🤖 AI Coding This Week
⏱ AI Coding Time: 76 hrs 30 mins (100.0%)
✍️ 43,948 lines written by AI, 2 lines written by hand (100.0% AI-written)
🔤 290,126,774 Input Tokens, 23,055,520 Output Tokens
💵 $1959.62 Estimated AI Cost This Week
🧠 1574 AI Sessions, 1965 AI Prompts
Opus 24,546 lines ██████████████░░░░░░░░░░░ 55.74 %
GPT 16,947 lines ██████████░░░░░░░░░░░░░░░ 38.48 %
Sonnet 1,970 lines █░░░░░░░░░░░░░░░░░░░░░░░░ 04.47 %
K 553 lines ░░░░░░░░░░░░░░░░░░░░░░░░░ 01.26 %
Haiku 24 lines ░░░░░░░░░░░░░░░░░░░░░░░░░ 00.05 %
🔎 AI Coding Insights:
🤖 AI-Driven — 100.0% of written lines came from AI
📚 Verbose Prompter — average 6,007 characters per prompt
🎯 One-Shot Prompter — average 1 prompts per session
🚀 High AI Trust — 0.0% of changed lines were hand-edited