Setting up a weekly reflective pass to extract rules where the agent corrected itself or I asked it to change something along with extracting repetitive things to skills have been really productive
48
Mukul retweeted
I expect this trend to really grow.
Multidisciplinary engineers, especially with high customer empathy and product sense will be in high demand
aboutamazon.com/news/company…
Mukul retweeted
The coding interview is dead, at least for my team at Anthropic. I've seen enough to know that it no longer provides signal. I removed it from loops this month.
I have a spicy, optimistic take - I think there's an opportunity for junior engineers: felixrieseberg.com/hunger-be…
Mukul retweeted
one of the most liberating realizations a human can have:
almost all of your thoughts are completely useless, wrong, and meaningless.
they are just biological feedback from a dysregulated nervous system, old childhood defense mechanisms, and genetic conditioning.
i have open sourced the repo for the Cricket 07 browser port (minus the game files)
link: github.com/Dhaiwat10/cricket…
tldr of how i did it:
• took the original 2006 x86 .exe and statically recompiled it to C (~20k functions, ~1M instructions)
• wrote a small windows api layer (heap, files, threads, registry, timers) so the recompiled code can run outside windows. got it booting natively on my mac first
• translated the game's direct3d 8 calls to webgpu, including the old dx8 vertex/pixel shaders -> wgsl
• compiled the whole thing to wasm with emscripten
• multiplayer runs on netplay over websockets with a cloudflare durable object per room
Crazy
Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine.
• Best-in-class fast local coding
• Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark
• Built from scratch in under 4 months
Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.
1
81
Mukul retweeted
Everyone is always rooting for you. Your parents want you to be a great son. Wife wants you to be a great husband. Your boss wants you to be a slam dunk hire. Every first date you’ve ever been on they’ve been rooting for you to get laid. Every time you started to tell a joke people hoped it would have a hilarious punch line. Your proximity to anyone is a reflection of themself, meaning the deck is never stacked against you, and your failures are completely your own
Readers added context they thought people might want to know
This post copies word-for-word a May 2026 post by @AdemLuz.
x.com/AdemLuz/status…
Mukul retweeted
i’ve spent my career in early bizops roles. here are 4 things that made me better at the job:
1/ get technical enough to unblock yourself. @austinh___ had me spend 6 months larping as a data engineer at ramp with 0 experience. solving problems without borrowing eng time is a superpower
2/ be friends with EVERYONE. organize happy hours, stay for late night karaoke, don’t eat lunch at your desk. early culture building is everyone’s job! knowing who to ask and feeling comfortable asking makes the whole team 10x more resourceful
3/ be annoying about getting clarity. being largely unmanaged doesn’t mean figuring everything out alone. push for answers, bump messages, challenge assumptions. founders are not exempt. i often corner mine in the office :)
4/ pick 1–2 problems at a time and know them cold. i’ve done post-sales, support, recruiting, growth… only ever well when i narrowed my focus. being needed everywhere is flattering, but a terrible prioritization strategy
interesting
a few of my predictions for the next 6 months of ai data:
- a monopolistic human data buy side layer emerges. right now you can use zap mail, heyreach, instantly and claude max plan to reach out to practically every potential human data buyer on the world in 7 days and (even with the worst strategy) at least come out with roughly 20 buyers spanning rl builders, neo labs and (if lucky) one or two major labs, who you could easily offload $20-$50m cumulatively from the very start. it is not going to be this easy to find buyers forever. as labs inevitably get f*cked by bad apples on rights, de-id, QC and the rest, a select few will emerge as “transaction gates” in which every corpora must flow them in order for the labs to deem as safe to buy. analogous to how a few labs already don’t touch corpora if it was de-id by Tonic Textual.
- the waterline continues to rise. ai labs are buying real company data to build artificial workspaces to teach their models how to do an office job. using a headcount as a metric, 3 months ago they bought company internals with 10+ headcount, now they almost exclusively purchase from companies with 50 or more employees. as models saturate these smaller workspaces, demand for smaller company corpora plummets whilst demand for the already dwindling supply of large sprawling company internals increases. yet doesn’t skyrocket as at a certain point if the price becomes too high then labs fall back on synthetic expert work. it goes without saying that small companies in a esoteric domain models underperform in that hold vertical depth (e.g a 30-man freight brokerage) will still hold immense value.
- reality gets graded more and more. as the most knowledgeable domain experts begin to fail in writing valid rubrics that aren’t instantly saturated we are going to have to get reality to play the verifier, domain expert and environment, if we want progress to continue. rubrics shift from the attributes of a perfect NDA to a lab measured critical tempurature.
- atom environments. why hasn’t the vertiginous increase of ai capabilities led to huge economic growth? how can we get non-tech people to understand how intelligent ai is already? how do we find the best value/token ratio? to “solve” these questions I believe need to give ai a “body” in which it can interact with the real world at extremely low bandwidth. giving ai a body sounds like a humanoid which is obviously just one small slice of real world interaction, a little bit like how literally none of the ways we interact with in the real world is only via our real body but via tools instead like handles, vehicles, instruments and cutlery. maybe I should reframe it from giving ai a body to giving the world ai
- protocols for company data emerge. for such a fuzzy yet crucial area of ai, it is insane how unorganized certain steps in company data brokerage are, such as giving context on what your companies internals actually hold.
40
Mukul retweeted
Meet TYGER: a decision model for speech that opens the way to full-duplex voice agents.
It listens to the mic and your TTS at once, and every 80ms decides whether to speak, yield, or chime in with an “mm-hm”.
No need for VAD anymore.
Works with any LLM and TTS.
huggingface.co/nineninesix/t…
this is such an important tool rn
HTML might be the new Markdown, as @trq212 pointed out early this year. But collaborating on HTML is still painful.
At @Razorpay, as agents started generating interactive docs and decks, we needed a way for humans and agents to collaborate on them in real time. So one engineer built SlashDocs. The rest of the team (and our AI engineer Slash) kept making it better.
Today, it has replaces google docs for 50%+ of all new docs at Razorpay, changing how humans and AI agents collaborate.
And now we're open-sourcing it!
We've spent years building tools for humans to collaborate. Now we need tools for humans and AI agents to collaborate.
52
Mukul retweeted
I’ve been fighting Narendra Modi and the RSS for over a decade now. I understand them like you would understand your opponent - inside out, like the back of my hand.
The core characteristic of these people is cowardice.
Narendra Modi, Amit Shah, Mohan Bhagwat, Gyanesh Kumar and Ajit Doval, the villains of this story, are slowly finding the space closing around them. We are closing that space.
Mr. Modi, do whatever you want. Call out all the police: 10,000, 20,000, 50,000, whatever you want.
Arrest all the leaders. Arrest all the young people. Arrest us all.
We are surrounding you, and we are not going to back down. Mark my words: we already have the information we need.
We are going to defend our Constitution. We are going to protect the voice of the Indian people. And we are going to ensure that what was done to India over the last 10-15 years will never, ever be done again.
Youth of India, I am proud of you. Keep protesting. We are standing with you.
We are going to ensure that this regime and the people who have attacked our Constitution are defeated.
मैं नरेंद्र मोदी, RSS और इनसे जुड़े लोगों के ख़िलाफ़ एक दशक से भी ज़्यादा समय से लड़ रहा हूँ। मैं अपने विरोधियों को उतनी अच्छी तरह समझता हूँ, जितना कोई भी अपने प्रतिद्वंद्वी को समझता है। मैं इन्हें अंदर तक जानता हूँ, इनकी हर चाल से वाक़िफ़ हूँ।
इन लोगों की सबसे बड़ी पहचान है - कायरता।
नरेंद्र मोदी, अमित शाह, मोहन भागवत, ज्ञानेश कुमार, अजीत डोभाल - इस कहानी के खलनायक - अब धीरे-धीरे घिरते जा रहे हैं। इनके चारों तरफ़ से दायरा सिमट रहा है।
जो करना है, कर लीजिए। जितनी पुलिस बुलानी है, बुला लीजिए - 10 हज़ार, 20 हज़ार, 50 हज़ार, जितनी चाहें बुला लीजिए।
सारे नेताओं को गिरफ़्तार कर लीजिए। सारे युवाओं को गिरफ़्तार कर लीजिए। हम सबको गिरफ़्तार कर लीजिए।
घबराइये मत - हम आपको चारों तरफ़ से घेर रहे हैं, और हम पीछे हटने वाले नहीं हैं।
मेरी बात याद रखिए - हमें जो जानकारी चाहिए, वह हमारे पास पहले से है।
हम अपने संविधान की रक्षा करेंगे। हम हिंदुस्तान की जनता की आवाज़ की हिफ़ाज़त करेंगे। और हम यह सुनिश्चित करेंगे कि पिछले 10-15 सालों में हिंदुस्तान के साथ जो हुआ है, वह दोबारा कभी, किसी भी सूरत में न हो।
हिंदुस्तान के युवाओं, मुझे आप पर गर्व है। अपना विरोध जारी रखिए। हम आपके साथ खड़े हैं।
हम यह सुनिश्चित करेंगे कि इस सरकार और हमारे संविधान पर हमला करने वालों को शिकस्त मिले।
Mukul retweeted
टू थाउसैंड सैंतालिस में
सुअर उड़ेंगे भाग जगेंगे
हीरे मोती पेड़ लगेंगे
पापी सब सॉरी बोलेंगे
ठग भी बिल्कुल नहीं ठगेंगे
नक़ली दूध ना नकली नेता
सब कुछ होगा खालिस में
टू थाउसैंड सैंतालिस में।
ट्रेन चलेगी गोली जैसी
हर दुश्मन ही मित्र भी होगा
पुलिस की गोली घाव करेगी
लेकिन उसमें इत्र भी होगा
अंदर से ख़ाली हो चाहे
सब चमकेगा पालिस में
टू थाउसैंड सैंतालिस में।
#2047
Mukul retweeted
At this point you have to believe that the Delhi Police secretly loves RG. Why else would you give him such visuals?
A really nice read.
Great post by @hezarfendd about using AI for learning new things
ogzhanolguncu.com/blog/how-t…
1
64
Mukul retweeted
My best career advice:
At every job you should either learn or earn. Either is fine. Both is best. But if it's neither, quit.






















