←
Arun Kant Sharma279 posts
Arun Kant Sharma avatar

Arun Kant Sharma

@arunkants

plural of "regex" is regrets | AI engineer

◷ Joined September 2011▧ 30 Photos and videos
adithyabhat avatar
⟳ Arun Kant Sharma retweeted
adithyabhat@adithyalbhat17·Nov 17

Scale sovereign and local AI, need of the hour for technological growth of the country. @smdcmc (@crane_vc ) and @siddharthaa7 (@neon_fund ) have backed my founder to do it. Exited to have @smdcmc (@crane_vc ) and @siddharthaa7 (@neon_fund ) for backing us.

We've raised $1.9m in preseed to scale sovereign and local AI at JohnAIC! We intend to use the funding to secure GPU and supply chain and to build amazing applications to run out of box on JohnAIC. Really greateful to @smdcmc (@crane_vc) and @siddharthaa7 (@neon_fund) for backing us!

Post attachment
Sasank Chilamkurthy avatar
⟳ Arun Kant Sharma retweeted
Sasank Chilamkurthy@sasank51·Nov 17

We've raised $1.9m in preseed to scale sovereign and local AI at JohnAIC! We intend to use the funding to secure GPU and supply chain and to build amazing applications to run out of box on JohnAIC. Really greateful to @smdcmc (@crane_vc) and @siddharthaa7 (@neon_fund) for backing us!

Post attachment
Tim Sweeney avatar
⟳ Arun Kant Sharma retweeted
Tim Sweeney@TimSweeneyEpic·Nov 3

How about there is a free market for stores, and they compete for consumers and developers on all facets of services and quality? This is how every other industry in the world works, except on iOS where there's a store monopoly developers must bow down to for mercy.

When I ran App Review, we talked about charging developers for reviews. We discussed a few versions: a cap on reviews per app, a VIP fast pass to the front of the queue, and paid expedite. None of it shipped. It would have looked like a money grab, and nobody wanted Apple’s name on that headline. That was years ago. Since then, three things have changed. New apps are up 84%. Commission rates are falling in the US, the EU, and China. And Apple has a new CEO who needs to make his mark. This year Apple has already started terminating developers for submitting too often, and making new developers prove themselves before anyone looks at their app. I think the next step is a price list for review, and I think I know what’s on it. Steve’s rule was that a human sees every app. I gave the first ground on that rule myself, and I wrote about how in the piece. phillipshoemaker.substack.co…

Sasank Chilamkurthy avatar
⟳ Arun Kant Sharma retweeted
Sasank Chilamkurthy@sasank51·Oct 29

First JOHNAIC just got manufactured in our new factory and there's plenty of space to scale the production!

Post attachmentPost attachment
RossRadio avatar
⟳ Arun Kant Sharma retweeted
RossRadio@cqcqcqdx·Oct 19

"What are you doing?" - "Oh just playing a Switch game."

0:58
Arun Kant Sharma avatar
Arun Kant Sharma@arunkants·Oct 18

deepseek cooking again 🔥🔥🔥

🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4

Post attachmentPost attachment
Sasank Chilamkurthy avatar
⟳ Arun Kant Sharma retweeted
Sasank Chilamkurthy@sasank51·Oct 7

What does it cost to rent an always-on 8×H200 cluster for a year? AWS (p5en.48xlarge), On-demand, Mumbai (on Oct 26): ~$660K/year And the cheapest AWS rate I found needs a 3-year commitment. At that point you've committed like an owner, without owning anything at the end.

DHH avatar
⟳ Arun Kant Sharma retweeted
DHH@dhh·Sep 20

The age of agents belongs to Linux!

Post attachment
DeepSeek avatar
⟳ Arun Kant Sharma retweeted
DeepSeek@deepseek_ai·Sep 10

💾 Smaller KV cache. Bigger savings. Compared with the previous generation, V4.1-Flash’s KV cache needs just: 🔹 1/4 the HBM 🔹 1/8 the SSD storage Cache-hit charges often account for a large share of agent costs. Compressing the cache cuts those costs significantly. 3/6

Post attachment
Arun Kant Sharma avatar
Arun Kant Sharma@arunkants·Sep 8

Is this real? 50x less compute? What about data?

Frontier pretraining is said to be a big-lab-only game. We don’t have 100k chips yet, so there’s only one way: algorithmic efficiency. Our new recipe matches DeepSeek V4 Pro’s pretrain using 50x less compute – that’s roughly half the FLOPs used for GPT3, or ~$0.5M on GB200. magic.dev/blog/pretraining

Post attachment
Arun Kant Sharma avatar
Arun Kant Sharma@arunkants·Sep 8

If this is true, it gives new meaning to "frontier lab" -> front-running your users research We need full clarity on what happened from both sides

this is insane. Levent and Tristan make major independent progress towards Navier–Stokes, with help from Claude and Codex. word reaches OpenAI. days later, OpenAI claims internal NS results, coincidentally using the same novel methods as Levent and Tristan. Tristan asks if OpenAI trained on their private Codex sessions. he never gets an answer. then OpenAI pressures Tristan to publish without Levent because Levent works at Anthropic. when Tristan refuses, he says OpenAI threatens his career.

Tuta avatar
⟳ Arun Kant Sharma retweeted
Tuta@TutaPrivacy·Aug 29

Google says YOU must wait 24 hours if you download an app onto your OWN DEVICE if it's from an "Unverified developer". What ever happened to user choice and Android freedom?! 😡 Google has started rolling out its new ""Advanced Flow", which requires you to wait a 24-hour waiting period if you want to install apps from unverified developers. Find out more: tuta.com/blog/android-side-l… Sign the petition: change.org/p/keep-android-op…

Post attachment
Arun Kant Sharma avatar
Arun Kant Sharma@arunkants·Aug 26

What timeline is this? Is Apple really going to save local and prosumer AI? Not Intel, AMD, Nvidia or Qcom