ojas

i crave knowledge and cats and borgirs || techbro, gamer, jack of not all trades || designing systems and ai agents

[ojas @bionic_BUG]

@stupidtechtakes been using zen since a few months, wdym its unusable?

[ojas @bionic_BUG]

why is spending unnecessary amounts of token a flex these days

[Vedant Kumar @itsvedantkumar]

yk youre down bad when you have to use gemini 😭

@claudeai i miss u ml

[Harshdeep Rapal @harshdeeprapal]

If only Indian government was as obsessed with AI as it is with Ethanol blending.

[ojas @bionic_BUG]

@schteppe thats a recipie for disaster, even skilled people can make mistakes. rather make the ci so that it skips the runs on noop changes

[Sandro @pupposandro]

I don’t care if the GPU drivers turn my setup into a CCP botnet.

If it makes NVIDIA lower prices for VRAM, I’m rooting for it.

[ojas @bionic_BUG]

@Bhavani_00007 ever heard of tmux os screen? even before ai agents, we have been using these for long compilations and stuff. this is so performative

[ojas @bionic_BUG]

@mil000 chinese farms would like to have a word

[anna @meowkoteeq]

we are making Flipper One (yes!!) and we'll make it your favourite little computer

but we need your help

[ojas @bionic_BUG]

@theo

[Md Ismail Šojal 🕷️ @0x0SojalSec]

Another bro,

[Socket @SocketSecurity]

Update: Socket has found 121 more compromised npm package artifacts across 84 package names, including 64 UiPath artifacts.

Combined w/ TanStack, the current known total is 205 affected npm package artifacts across enterprise automation, AI/MCP, auth, workflow, and dev tooling.

[ojas @bionic_BUG]

@kirat_tw yall turned gsoc into another rat race fml

[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.

[ojas @bionic_BUG]

we see shit like this, and then there are non technical people pushing to production, what can go wrong?

[Sebastian Aaltonen @SebAaltonen]

You have to review all LLM code!

Codex 5.5 tried to push this awful hack to our Metal backend when it was coding font rendering. It decided to implement hacky "robust buffer access" style OOM check inside the shader and hacked our whole Metal binding architecture to add a special bind group slot 30 (hardcoded) to deliver sizes of all buffer bindings. This of course made the binding model super slow and required extra data for each buffer.

[ojas @bionic_BUG]

@naval wat

[luthira @luthiraabeykoon]

We implemented @karpathy 's MicroGPT fully on FPGA fabric.

No GPU.
No PyTorch.
No CPU inference loop.

Just a transformer burned into hardware, generating 50,000+ tokens/sec.

The model is small, but the idea is not: inference does not have to live only in software 👇

[ojas @bionic_BUG]

friendship ended with vscode, zed is my best friend now

[ojas @bionic_BUG]

@jeremie_strand true

[ojas @bionic_BUG]

2026 is the year of supply chain attacks

[Socket @SocketSecurity]

🚨 We’ve confirmed the [email protected] was compromised in the ongoing Mini Shai-Hulud worm attack.

The npm package includes a malicious preinstall hook that downloads and executes an unverified Bun binary, then runs an 11.7 MB obfuscated payload designed to steal Kubernetes, Vault, cloud, GitHub, and CI/CD secrets.

The attack closely overlaps with the SAP CAP, Cloud MTA, and [email protected] compromises.

[ojas @bionic_BUG]

chat jipiti

[✨ Celestia ✨ @WhereIsCelestia]

This got posted in a discord server I was in, and it absolutely sent me.