Vinoth Kumar

Director of Engineering at Refold AI, AI/Security

[Vinoth Kumar @vinothsparrow]

I built an operating system where the AI agent IS the process.

No Linux. No POSIX. A tiny LLM runs on bare metal in Rust — and a capability ABI means it can plan anything, but only do what it was granted.

It boots, browses the web, plays video, and talks. 🧵

[Vinoth Kumar @vinothsparrow]

@dhh Yesterday, I added Freedoom to ChittiOS https://t.co/fCIZYF9yqd
🤔

[Vinoth Kumar @vinothsparrow]

@dhh @dhh can you build Omarchy on top of this instead of Linux 😁

[Vinoth Kumar @vinothsparrow]

@dhh Just launched my OS 😅
https://t.co/jo5Q1JQKBA

[Vinoth Kumar @vinothsparrow]

I built an operating system where the AI agent IS the process.

No Linux. No POSIX. A tiny LLM runs on bare metal in Rust — and a capability ABI means it can plan anything, but only do what it was granted.

It boots, browses the web, plays video, and talks. 🧵

[Vinoth Kumar @vinothsparrow]

~3,800 lines of mechanism in a ~300k-line kernel, so the part that has to be right is readable in an afternoon.

Research OS — not stable; run it in a VM. Apache-2.0, and I'd love contributors.

⭐ https://t.co/r4JPEhDVUf
📄 https://t.co/d4xlGk4hnc

[Vinoth Kumar @vinothsparrow]

I built an operating system where the AI agent IS the process.

No Linux. No POSIX. A tiny LLM runs on bare metal in Rust — and a capability ABI means it can plan anything, but only do what it was granted.

It boots, browses the web, plays video, and talks. 🧵

[Vinoth Kumar @vinothsparrow]

The honest part: the defence costs something. 25% of legitimate irreversible operations get refused because the turn touched untrusted data.

I built the cheap fix — per-value string matching — and it permits 4 attacks while recovering zero. The model is the dataflow.

[Vinoth Kumar @vinothsparrow]

@GoogleAIStudio https://t.co/ECQ70Ioe2U this hobby Agentic OS

[Vinoth Kumar @vinothsparrow]

Just got the green light for Anthropic��s Cyber Verification Program! 🟢. Good to be back. 😎 🎯 #bugbounty #cybersecurity #AI

[Vinoth Kumar @vinothsparrow]

🤯🤯

[Alexander Whedon @alex_whedon]

Introducing SubQ - a major breakthrough in LLM intelligence.

It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA),

And the first frontier model with a 12 million token context window which is:

- 52x faster than FlashAttention at 1MM tokens
- Less than 5% the cost of Opus

Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention).

Only a small fraction actually matter.

@subquadratic finds and focuses only on the ones that do.

That's nearly 1,000x less compute and a new way for LLMs to scale.

[Vinoth Kumar @vinothsparrow]

@sama 🤔🤔🤔

[Vinoth Kumar @vinothsparrow]

If you know, you know 👀

[Vinoth Kumar @vinothsparrow]

Zed + Claude Code — just wow 🤯.
Didn’t expect this combo to feel this smooth.
@zeddotdev @claudeai

[Vinoth Kumar @vinothsparrow]

Been on Cursor for a while, but Go dev never felt quite right with VS Code/Cursor.
Always had to fall back to GoLand.
Zed + Claude though… this finally clicks 👀