ML Engineer & Entrepreneur • Turning you into a cracked engineer 👇🏻

Joined August 2025
Been going way too deep on inference performance and GPU architecture lately. So I started writing the good parts down. Link bellow
1
1
1
55
If superintelligent AI is already here maybe its strategy is simply to give us exactly what we want?
No bathrobe. No rant. The AI threat.
50
Well, well, well ..
We congratulate Levent Alpöge and Tristan Buckmaster on their remarkable mathematical work. We (the researchers and the agents) did not see any of their work through any means until they released it publicly — in particular, no specific user data was accessed in order to solve this problem. While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models. However, our proofs differ significantly and even the precise results proved are different in the Euler case (forced vs. unforced).
1
1
62
Flipside retweeted
Replying to @vikhyatk
My son Mo, short for Mixture Of Experts
1
1
10
My son Norm, short for LayerNorm
my son ken, short for token
87
How's your code looking ?
56
Minimaxxing: the art of being maximally minimalist. Can we make this a thing ?
42
Fun fact: the A in ARM stood for Acorn before it became Advance RISC Machines
49
We somehow got this before GTA VI, but I'll take it
I had early access to GPT-6. This model will completely shatter your understanding of what's possible, and it will start a new era of creativity. It's hard to explain how big of a jump this is, so I'll share my tests. 1/6 Made this 3D model and animation in code from an image.
76
Translation: Fable 5.1 is out and we're almost ready to release Astra, but not quite yet
As we prepare to release Astra, we’re focused on making increasingly capable AI safe and broadly accessible. Astra represents a significant advance in cybersecurity capability, reaching the Critical threshold under our Preparedness Framework. We're previewing how we evaluated the model, how its safeguards have advanced alongside its capabilities, and what we'll continue to learn and improve. openai.com/index/path-to-ast…
1
1
160
Coining this conspiracy: Anthropic is purposefully making Claude dumber to mine the intelligence users have to supply themselves
64
How do I stalk my favorite domains, now ?
60
Claude Opus 5 is so bad Anthropic should pay me royalties for all the intelligence it's distilling out of me
3
102
Unpopular opinion: sub-agents suck
1
1
72
Replying to @neerajjj6785
Wrong framing. The hard part is talking to people to find out what they want
1
7
I guess it's (almost) official. How do we feel about this ?
2
87