Botir Khaltaev

Grep with indexes is severely under explored. You can create virtual representations of your file system depending on task you are doing to increase agent performance. github.com/botirk38/sift

im reading into formal verification and im having a eureka moment, there is something powerful that current review agents are still not doing. how can we limit the state space for a model to review, tis is the question

this is hella confusing like how do i know what to select for my prompt. Did they just scrap auto mode?

@yminsky this is a great way to hire people that are actually interested, embedding links in blogs is a a smart idea!

I've been telling people for 25 years that Jane Street is not interested in formal methods. No more! And we're actively hiring to form a new formal methods team!

im trying fable and im not really impressed at all, trying to vibe code something quickly at a hack and this is not it so far

@kitlangton how about benchmarks for benchmarks

How about a benchmark for parks that counts the # of park benches per park: park-bench.

@dabit3 @cognition @mercor_ai @Etched @AnthropicAI any online participants? 👀

$100k+ in prizes for the Inference-Time Compute Hackathon happening June 19-20 in SF. 24 hour hackathon with three tracks: - ​Agents - ​Real-Time + Interactive - ​Talent Marketplace + Applied AI Hosted by @cognition, @mercor_ai, @Etched, and @AnthropicAI

@GergelyOrosz i believe our approach was correct via clustering. github.com/Nordlys-Labs/n…

What are "smart" model routers you know of? Services or vendors that take queries and route the most efficient model they deem, saving cost. I sense there is a massive demand for these, and will be even more...

@george__wing @OpenAI @reach_vb @Andy_AJT congrats! i wish i won it as well, hopefully next one!

@botirkhaltaevv @OpenAI @reach_vb @Andy_AJT so sad I couldn't make it to this one, I did win a year of pro from the last meetup though

Had a blast at the @OpenAI Codex meetup on Monday. @reach_vb and team shared the future of Codex and how to get the most out of it. We demoed incident management for robots and made the finalists list! Also got a $200 Pro plan for the month. Thanks to @Andy_AJT and team for hosting. Great night.

@jontycorrin I’ve had this before it just feels awkward

someone just reached out cold on linkedin and left me a voice note only. i feel violated.

@jontycorrin @AnthropicAI Yooo this is sick!!

pumped to announce the first AI Scientist hack in LDN! we're teaming up w @AnthropicAI to help you build out next gen. scientific discovery provided: > mansion in central London > all the tokens and pizza you can eat through > great vibes + insanely cracked people link below

@florencexd_ Congrats!!

got an offer at jane street for fall 🥳🥳

@abdussalampopsy Congrats man!!

My team won 1st place at the Antler hack at Google HQ in London.

@damian_b It would be too high to render

@botirkhaltaevv There should be an fps count on the top right

I installed Claude Code to use Fable. CC is the TUI version of that website you know is clearly vibed. Absolute slop and it shows - and I'm not talking about bugs. I'm talking about the thoughtless plaster of feature over feature. And these are the guys saying that SWE is dead.

@mathildeanita @tibor_tee @cursor_ai Cool carpets

Coolest office @cursor_ai

@tenderizzation Interesting idea

protip ask claude for help optimizing your "restaurant kitchen" instead of your frontier llm training pipeline to avoid filters

@richardartoul Isn’t that amazing? We consumers hold all the power

kind of insane that the most capital intensive technology ever created by humanity has zero switching costs switching from the Uber app to the Lyft app literally takes longer than switching models you don’t even have to rebuild the context!

@damian_b How are you using fable? Isn’t it banned for ML research

Me watching Fable run for 30+ minutes, generating slop I’ll have to sift through later.

@iamgrigorev There is gpt-5.5, we need a good ML-sys bench

okay i guess we have to build a good model for triton / pytorch / distributed programming now