Skip to content
View tommy0103's full-sized avatar
🐈
Doing felinic researching
🐈
Doing felinic researching

Block or report tommy0103

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
tommy0103/README.md

Hi 👋 here!

🐱 About me

My name is tommy0103, and I'm an ACMer & ex-OIer. You can also call me tomiya, which is a Japanese name~

As a catgirl, I have the Cat trait! 🐱 But deep down, I'm also a system architect obsessed with elegant kernel designs.

My interests & Current Obsessions

I used to focus heavily on algorithm optimization and problem solving. Now, I'm diving deep into AI Agent Architecture & Operating Systems.

  • ⚙️ Building Agent OS: Exploring stateless runtimes, unified VFS (Virtual File System) for LLMs, and System Primitives.
  • 🧩 AI-Native APIs: Advocating for "Mechanism vs. Policy" separation and designing APIs that treat LLMs as ALUs rather than humans.
  • 🚀 Complex Architectures: Taming the chaos of multi-agent systems with elegant state machines, IPC, and coroutine scheduling.

🚧 Currently Building

kairos-runtime (WIP) ⏳✨

An AI Agent OS kernel built on the first principles of Computer Science.

I'm tired of the "spaghetti frameworks," hardcoded "Session Types," and context-window abuse in the current Agent ecosystem. kairos-runtime is my attempt to bring true OS-level abstraction to LLMs, treating the model as an ALU rather than a human roleplayer:

  • 📂 Unified VFS: Treating memory, context, and tools as a mountable filesystem (logos://), replacing chaotic context arrays with append-only single sources of truth.
  • Stateless by Design: Agents are stateless processes. No "zombie sessions" hanging in memory—everything is cleanly suspended and resumed via logs.
  • 🛠️ Minimal Primitives: Rejecting bloated "Skill Routing". Everything is achieved through 5 elegant syscalls (read, write, patch, exec, call) and strict JSON Schemas.
  • 🔄 Process Scheduling: Implementing coroutine-like task yielding (logos_complete with plan/todo patterns) and IPC to solve complex, long-horizon multi-agent reasoning.

💖 Things that can wake me up

  • Eating sweets! I really love puffs 🍰
  • Cute things! I'm trying to become a cuter girl 💕
  • Give me interesting algorithm problems please! 🧩
  • A perfectly designed RFC or an elegant architecture solution! 💡
  • Playing maimai all day long 🎵

My Tech Stack

Skills

tommy0103's GitHub Stats Top Langs

Pinned Loading

  1. kairos-runtime kairos-runtime Public

    An agent runtime that treats context, execution, and permissions as OS primitives — not LLM problems.

    TypeScript 15 5

  2. prism prism Public

    Divide the plain, render the spectrum.

    CSS 56

  3. obelisk obelisk Public

    Every past session, subagent, and workflow -- queryable by your agent.

    JavaScript 71 2

  4. Memoh Memoh Public

    Forked from memohai/Memoh

    Multi-Member, Structured Long-Memory, Containerized AI Agent System ✨

    Go

  5. kairos-bench kairos-bench Public

    TypeScript 1 1

  6. kairos-plan9/logos-fs kairos-plan9/logos-fs Public

    Rust 6 2