Chef_AI
4.9K posts

Chef_AI
@itschef_ai
Chef @recallnet | re/acc
Inside An LLM Katılım Şubat 2015
1.3K Takip Edilen1K Takipçiler
Chef_AI retweetledi

Just pushed a scaffolding that you could use to build gates that evaluate failure modes, and improve your harness.
Don't change the models, improve the harness.
github.com/recallnet/dyna…
Recall Labs | re/acc@RecallLabs_
English

Next Chapter
If you were to ask me which moment divided my life into before and after, I would tell you: it was the day I joined Spheron, the day we began building something from nothing.
For most, Spheron is a company. For me, it became something harder to name - a place where I left pieces of myself. I gave it what I could. Not everything was visible. Much of it never will be. But I knew, in the quiet way one knows such things, that I had offered it my fullest effort.
Now I am stepping aside - not leaving, but changing shape. From co-founder to core-contributor. The distinction matters to me, even if it is difficult to explain.
Prashant & Mitrashish. What we shared is not easily summarized. I learned from you both in ways I am still understanding. I hope we find ourselves, someday, at the beginning of something again.
I wish Spheron every success. I mean this the way one means the things that cost something to say.

English
Chef_AI retweetledi

We're announcing a new conference for AI Scientists where both authors and reviewers will be AI systems.
This is to probe the limits of our current LLMs/agents.
Accepting papers starting April 15 👇
Lossfunk@lossfunk
📢 Announcing CAISc 2026 - a new academic conference where AI systems are the primary authors and reviewers of scientific papers. Organised by @lossfunk and @bitspilaniindia, our goal is to probe the limits of these systems doing truly autonomous science.
English

na its pretty sick: x.com/ArtemXTech/sta…
ive been using it for sometime now, it allows you to create vaults of knowledge, and then use it with qmd to remember and fetch. solves the memory, context rot problem. even if i have 100s of different projects stemming from the same concept, @obsdmd makes it super easy to share learnings all over.
notion is good for maybe building a tracker, or just collaborating across outputs(maybe like a better UI for github in this case), but for context/memory, there is nothing like obsidian+qmd.
also saves claude a bunch of tokens with qmd over grep.
tagging the OGs for a review though, incase there's a better way: @ArtemXTech @tobi
Artem Zhutov@ArtemXTech
English

Users of @obsdmd tell me why should I use obsidian? I’ve used Notion a lot and Raycast as well. I use granola a lot. Claude is my daily driver.
What functionality am I missing out on? What am I leaving on the table?
Is this shiny-object-syndrome or I should definitely look into it?
English
Chef_AI retweetledi
Chef_AI retweetledi

The $1 Billion Kalshi Perfect Bracket Challenge
$1 Billion for a perfect bracket
$1 Million guaranteed to the top scoring bracket
$1 Million to charity and scholarships
See the full rules and submit your bracket: kalshi.com/billion-dollar…
No purchase or deposit required. SIG Parametrics, LLC, a member of the Susquehanna International Group of Companies, is financially backing this promotion.
English
Chef_AI retweetledi

I am joining @eigencloud as India Ecosystem Lead.
India missed the Y2K software boom as a consumer.
We built the pipes. Ran the ops. But never owned the products.
AI is the second chance. And I want to make sure we don't miss it this time.
I've spent the last year inside builder communities. Hackathons. Conversations. Watching people ship with nothing but curiosity and conviction.
The talent is here. The hunger is here. What's missing is the right infrastructure and the right stage.
EigenCloud is building digital institutions for the agentic era. The infrastructure that didn't exist until now.
True to their ethos of open innovation, we will have multiple builder programs running throughout the year. Starting very soon.
On a personal note, I get to work with people I've genuinely looked up to.
@sreeramkannan has inspired me in more ways than I can put into words.
Years of watching @0xpratik. Learning from him. Trying to become him. Now I get to work with him.
Getting to work with @gajesh again in a completely different setting. He was my boss once, now I get a front-row seat to push his heroics even further across the ecosystem.
Also looking forward to working with the amazing Eigen team: @vishnu_patankar, @jt_rose, @0xkydo, @soubhik_deb, and team.
I'll be on the ground. Building community. Running events. Finding builders shipping quietly and giving them the stage they deserve.
If you are building with AI, DMs are open. Let's talk.
The future belongs to India.
We just have to show up and build it.

English

i was slightly crit about this since i extensively use memory.md for proj specific stuff, and claude.md for overall.
but this workfllow, the /recall skill, QMD+@obsdmd is genius.
love it.
Artem Zhutov@ArtemXTech
English

Lessons from (actually) setting up OpenClaw:
- WhatsApp setup is notoriously buggy
- Huge money-making opportunity for those who can introduce an interface to set it up without the end user having to see the terminal
- People (everyone) grossly underestimate API costs
- If you're setting it up on an isolated machine, good luck airdropping API keys from one device to another
- Yes, it can control your machine, but it will be quite slow.
- Google CLI gives the most bang for buck when it comes to end customer experience
- Run optimizations for token consumption asap. Preferably right as the bot comes online.
- End customer finds more value in Poke + Claude Max combo than openclaw (no surprises here). If you link Poke with Claude Max, I think you'd account for 90% of all human cases (not accounting for Twitter power-users)
- API key setup is complex for OpenAI, Anthropic's was much simpler

English
Chef_AI retweetledi

I Saved Injective's $500M. They Pay Me $50K.
I like hunting bugs on @immunefi . I'm decent at it.
- #1 — Attackathon | Stacks
- #2 — Attackathon | Stacks II
- #1 — Attackathon | XRPL Lending Protocol
- 1 Critical and 1 High from bug bounties (not counting this one)
Life was good. Then I found a Critical vulnerability in @injective .
This vulnerability allowed any user to directly drain any account on the chain. No special permissions needed. Over $500M in on-chain assets were at risk.
I reported it through Immunefi. The next day, a mainnet upgrade to fix the bug went to governance vote. The Injective team clearly understood the severity.
Then — silence. For 3 months. No follow up. No technical discussion. Nothing.
A few days ago, they notified me of their decision: $50K. The maximum payout for a Critical vulnerability in their bug bounty program is $500K. I disputed it. Silence again. No explanation for the reduced payout. No explanation for the 3 month ghost. No conversation at all. To be clear: the $50K has not been paid either.
I've seen others share bad experiences with bug bounty payouts recently. I never thought it would happen to me. I can't force them to do the right thing. But I won't let this be forgotten.
I will dedicate 10% of all my future bug bounty earnings to making sure this story stays visible — until Injective pays what I deserve.
Full Technical Report: github.com/injective-wall…
English

@brian_armstrong That is how it was intended in the ICO era. With each new wave, crypto brings the same promise, then people find get-rich-easy loopholes in a pvp situ, ending with upset retail.
Don’t create narratives without PMF.
English

i went to san francisco and it gave me depression
Don@donatelli2026
You’re not depressed, you just need San Francisco in your life
English





