21 • Engineer • Developer • Building Kenesis 👓

Joined November 2022
239 Photos and videos
Pinned Tweet
27 Jul 2025
If I ever create something half as creative as Spider-Verse, I’ve won.
3
24
2,877
wth is going on
JUST IN: Andrej Karpathy, a top AI scientist at Anthropic, is reportedly barred from accessing the company’s most advanced AI model because he is not a U.S. citizen.
11
Jun 12
🐐🐐
24 Dec 2011
Kanye stopped by the SpaceX rocket factory today.
13
Jun 12
my life should sound like the fashion killa song by asap rocky
41
Jun 12
New SpaceX IPO like animation is really cool
2
4
889
15,979
Jun 11
Dylan throughout this episode
Replying to @foundersfund
MAFIA EP 002
41
Jun 11
my recent favourite show is on
Replying to @foundersfund
MAFIA EP 002
15
Amrish retweeted
Jun 10
.@innercircle_so started with just a few of us, building things, becoming friends. 2.5 years of just walking towards whatever felt right got us here. a real space for builders to come build together. t-10 to open. indiranagar. if you're a solo founder or a team building in silo and want to work with us, comment inner circle to get in early.
116
22
298
40,500
Jun 9
aws ftw
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use. Its capabilities exceed those of any model we’ve ever made generally available.
14
Jun 8
Chennai had The Odyssey IMAX bookings disappear within minutes because we only have 2 IMAX screens. Bengaluru has 6 IMAX screens, opened earlier, and still has better availability. The demand in Chennai is clearly there. @IMAX , please bring more IMAX screens here.
3
142
Jun 8
dei @district_india eppoda chennai la Odyssey booking open pannuve saala blr, bom la laam open pantaan
1
1
408
Jun 5
this is like love island for founders
We put 12 of our favorite people in a room @sama @zoink @bryan_johnson @PalmerLuckey @cyantist @traestephens And made them play the diabolical game of deception, Mafia And filmed it all live with @micsolana as the moderator Good fun :)
82
Jun 3
the icon pack that @Google uses since its I/O 2026 is really good.
1
235
Jun 3
an indian and a pakistani shipping together. somewhere, the internet just got a little more peaceful.
Introducing @heyclicky teammate #1. Welcome, Nilesh. If you want to help build the the next AI interface that millions use everyday, definitely reach out. I'm hiring. Engineering, film, design, everything.
15
3
218
26,351
May 30
i wish i could talk to @boardyai whenever i want on a call the natural conversation ability of it, without pushing anything is really great.
80
Amrish retweeted
build something wonderful is OFFICIALLY on! banger kick off session by @akhil_bvs & @RajansheeS 🫡 had an amazing time meeting everyone and sharing so many laughs alreadyyy
5
3
46
1,901
May 28
me watching the gap between “i have an idea” and “i shipped it” collapse in real time i used to pray for times like this
May 28
Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors. Available today at the same price.
56
Amrish retweeted
Meet the Build something wonderful squad 🌸
been such a dream to finally make this happen with @OpenAIDevs codex. introducing : buildsomethingwonderful.com because it's time we stopped waiting for permission. come build : something useful something beautiful and real. 15 solo founders. 4 weeks to build, launch, and validate a real product. show the world something wonderful and unlock upto $20,000 in API credits. applications open now.
5
7
49
16,045
May 26
"I'll update the README.md with the corrections" my ass
43
May 26
hot take: the best ai model is not always the biggest one. tencent’s new hy-mt2 translation models are a good example. on multiple translation benchmarks, they seem to beat or match much larger general models from gpt, gemini, deepseek, qwen, and others. the interesting part is not “hy-mt2 is better than all of them.” it’s not. the interesting part is that for one specific job, translation, a focused model can be cheaper, faster, easier to deploy, and sometimes more accurate than using a giant frontier model for everything. the 1.8b version can reportedly be compressed to around 440mb. that changes the deployment game. cloud is not always the answer. bigger is not always the moat. general intelligence is not always the product. the future might be a stack of specialized models doing specific jobs insanely well.
🚀 Open-source upgrade unlocked. Tencent Hy-MT2 is now under Apache License 2.0 — maximum freedom for research, commercial use, fine-tuning, and derivatives. No strings attached.😎😎😎 Proud to push model weights back to the community. Our two variants are currently sitting at #1 and #4 on the Hugging Face trending leaderboard. Clone, fork, break things, ship feedback. The iteration loop is live.🔥 Let’s keep building the frontier together. #Tencent #Hy #HyMT2 #Apache2 #HuggingFace #OpenSourceA
1
89
May 26
I love how people get excited for every single model release, only to settle down with sonnet 4.6
31