class Ty:
role = "AI Agent Developer"
focus = ["agent platforms", "rust systems", "automation pipelines"]
stack = ["Python", "Rust", "TypeScript", "Go"]
building = "arcana — controllable, reproducible, evaluable agents"
shipped = ["AI_novel", "arcana", "tachyon", "messageinfra"]
believes = "agents should do real work, not just chat"|
Controllable · Reproducible · Evaluable Agent Platform A Python agent platform built for real production use — not demos. Reproducible runs, evaluable outputs, and controllable tool surfaces.
|
小说 → 短视频 一键自动化 AI 小说推文自动化工具，把小说转换成有声书 + AI 配图的短视频，适配抖音/小红书。
|
|
Order Matching Engine in Rust 52ns median latency · 8M+ ops/sec. A lock-free, high-throughput matching engine for serious market infrastructure.
|
Learning Rust the hard way Notes and experiments as I dig into Rust's type system, ownership model, and systems-level patterns.
|
- 🛠 shipping —
arcanaagent platform +messageinfrafor information workflows- 🧪 contributing to — NousResearch/hermes-agent agent framework
- 📚 learning — lock-free data structures & agent evaluation benchmarks
- 💭 thinking about — how to make agents that are actually reproducible