Pavle Padjin

Cofounder of Hoid ( CS @ Cambridge

[Pavle Padjin @ppadjin]

@elliotarledge nice

[Pavle Padjin @ppadjin]

@bcherny @scaling01 really? my personal feeling is that you need to micromanage it to get the result, unlike fable

[Pavle Padjin @ppadjin]

If some parts are very performance critical and bug prone, it is probably good ROI to manually inspect it. In any case, you should have an agentic workflow for reviewing upon making a PR

[Pavle Padjin @ppadjin]

I don't really understand people saying "never manually review your code" or "always manually review your code". It is not binary, it depends what part of codebase you're editing

[Pavle Padjin @ppadjin]

pushing FLOP/s through the roof(line). check out the blogpost

[Momcilo Mrkaic @momcilo02]

Our first blogpost: @hoid_ai 's bet that traditional AI Compilers need to be redesigned from first principles.

We’re also publishing our initial results. Enjoy!

https://t.co/QudqWzmNdH

[Pavle Padjin @ppadjin]

@__tinygrad__ @bingxu_ @Etched I understand the critic for the lack of transparency, but I have hard time believing these big shots would market the company if the tech is bullshit. Even if they get stock for free.

[Pavle Padjin @ppadjin]

@Johnbms6n @nicholasnlawton Elon's companies are VC funded

[Pavle Padjin @ppadjin]

You just know they were waiting for another Github outage to publish this. They probably didn't have to wait a long time tho

[Cursor @cursor_ai]

Origin, our code hosting platform, is now live.

It's fast, easy to use, and deeply integrated with Cursor.

Get started by syncing your repos from GitHub.

[Pavle Padjin @ppadjin]

@thsottiaux I use claude code harness which i prefer to codex (more features) and also I prefer fable to sol. opus 5 is a necessary evil

[Pavle Padjin @ppadjin]

@__tinygrad__ tinycorp is very sneakily becoming a hardware company

[Pavle Padjin @ppadjin]

@paulg a step further, fast mode in cc/codex is a genius way to increase margins and revenue.
faster tokens = more tokens spent = higher revenue.
fast tokens = premium price per token = higher margins.

[Pavle Padjin @ppadjin]

@exolabs @NVIDIAAI concurrent_users = 2

[Pavle Padjin @ppadjin]

@venkat_systems that's why AMD is partnering with Cerebras

[Pavle Padjin @ppadjin]

@PatrickToulme I am also wondering what is a good interface for agents. It would need to draw some useful insights that are hard to find with raw session transcript in jsonl file

[Pavle Padjin @ppadjin]

@thsottiaux speed of shipping new features, it always feels ~3-4 weeks behind cc. I think pi has a good marketplace approach to this so community just builds them in a day

[Pavle Padjin @ppadjin]

@maximumagi @momcilo02 @Zelja__ @speedrun @hoid_ai yeah, wonder what gave it away hahaha

[Pavle Padjin @ppadjin]

We're coming out of stealth, it's time to build something great!

[Momcilo Mrkaic @momcilo02]

tl;dr of life recently: @ppadjin , @Zelja__ and I quit our jobs to build a startup (link in comments) and we're backed by @speedrun . It is the best time in history to build a company and we didn't plan on missing the boat.

@hoid_ai is an agentic AI compiler that turns any AI model into a highly optimized implementation for the given hardware.

We’re building the layer between AI models and the silicon they run on, using AI itself to explore the optimization space and squeeze out every last bit of performance.

If you're in the AI inference space hit me up let's chat.

[Momcilo Mrkaic @momcilo02]

tl;dr of life recently: @ppadjin , @Zelja__ and I quit our jobs to build a startup (link in comments) and we're backed by @speedrun . It is the best time in history to build a company and we didn't plan on missing the boat.

@hoid_ai is an agentic AI compiler that turns any AI model into a highly optimized implementation for the given hardware.

We’re building the layer between AI models and the silicon they run on, using AI itself to explore the optimization space and squeeze out every last bit of performance.

If you're in the AI inference space hit me up let's chat.

[Pavle Padjin @ppadjin]

@TanayKmr @opencode I think most OSS projects are currently fighting the pile of slop PRs. It is probably as frustrating for them as it is for respectable contributors.