Sudarshan Shah

I am Software Engineer at IBM, working in Java/Angular tech stack.
I enjoy to learn and explore new tech stack and ecosystem!

[Sudarshan Shah @sudarshandev119]

Microsoft doing some real good stuff.
Keep going and make more open weight models!

[Mustafa Suleyman @mustafasuleyman]

We now have state-of-the-art pre-trained base models in the 3T-parameter weight class, built entirely in-house at MAI.

Post-training results coming soon -- stay tuned!

Lots of details in our technical blog post.

https://t.co/xcZ5uV4TAp

[Sudarshan Shah @sudarshandev119]

Dev experience with Typescript:

NextJS: 5/10
Tanstack Start: 20/10

Developer retention:

NextJS: maybe 50%
Tanstack Start: 100%

This is just the start (no pun intended, maybe), with time, more people will appreciate how good is Tanstack Start!

[Marius @mariusdotdev]

@nextjs vs @tan_stack

[Sudarshan Shah @sudarshandev119]

@rishikagupta__ Yeah, I am thinking about someday being a billionaire😎😇
Good motivation👍

[Sudarshan Shah @sudarshandev119]

@mehulmpt I do use Bob daily at IBM, and I must say it's actually a really nice tool. It's gets the work done,and even the complex ones too.

I have been building all new features and modules in our project using Bob, and I just review the code and test the features.

[Sudarshan Shah @sudarshandev119]

@techie_piyush I thought AI agent broke your Sandbox to place an order🤭

Anyways, this is really a good choice.
Unboxing video upcoming🤔

[Sudarshan Shah @sudarshandev119]

@mehulmpt Rather then targeting Maths as new benchmark to prove themselves, this should be the new benchmark for AI model.
It would be very interesting😅

[Sudarshan Shah @sudarshandev119]

@Hiteshdotcom Jaisa ki aapne kaha tha: jab tak inka IPO nhi aata, Yeh aise hee darate rahenge.

[Sudarshan Shah @sudarshandev119]

@mehulmpt Oh yeah...but still I am using Opus 5.5 on Pro plan without hitting limits and enjoying new models release.

What a great time to be Software Engineer🙂

[Sudarshan Shah @sudarshandev119]

@mehulmpt But still if you think about it, Meta was no where in the race until it suddenly closed up with top frontier labs and that too by releasing SOTA open weight models.

AI race getting interesting, and pace should accelerate more.

Interesting times ahead!!!!

[Sudarshan Shah @sudarshandev119]

@nirudhuuu Chai toh app bhul hee gaye...aise kaise kaam chalega☕😅

[Sudarshan Shah @sudarshandev119]

@tannerlinsley Congratulations! This is the best & refreshing thing to happen in tech space when everyday news is all about new AI model & tool release.

Its really a big moment for you and Tanstack team, and the article you wrote is impressive. I read word to word, no AI summary😅.

[Sudarshan Shah @sudarshandev119]

I am personally very excited about GA.
I learned this a few weeks ago & it's so amazing, I am in love with it.

The best thing to happen to JS/TS ecosystem in Tanstack stack a whole.

Thanks @tannerlinsley and whole Tanstack team for making this with love and passion🙏 ♥️

[TANSTACK @tan_stack]

TanStack Start 1.0 is officially stable! 🎉

A full-stack web framework for React and Solid, already used to build millions of apps, from Fortune 500 companies to your favorite tools and weekend projects.

https://t.co/DA8E8cWnBU

[Sudarshan Shah @sudarshandev119]

@AlemTuzlak I wish Tanstack Start become stable finally.
No more NextJS to write.
oh..Anyways I am not writing NextJS anymore, it's either Tanstack Start with Rust or Spring Boot😉😎

[Sudarshan Shah @sudarshandev119]

I really admire your tech content on YouTube @mehulmpt, but putting @Hiteshdotcom sir on thumbnail with the kind of title you used is not really good😐.

Unless you both have a understanding about it, I would say it's wrong to point to a genuine & good teacher.

[Sudarshan Shah @sudarshandev119]

Best Employee of the year 2026 award is definitely going to @alexandr_wang for sure at Meta.

Blockbuster product release Muse, and beating down every frontier lab including @Google!

I will be trying it out soon...and I am very excited for it🔥

[Alexandr Wang @alexandr_wang]

MUSE WONDER

this is the culmination of over a year of research and infrastructure work in msl

since we originally shipped muse spark in april, we have continued to scale our models across pre-training and RL, now reaching the frontier 🏆

see our model's improvements on the artificial analysis index:

[Sudarshan Shah @sudarshandev119]

People were saying @finkd wasting money on AI like he did on Metaverse.

But it seems like he really made a good business decision to hire @alexandr_wang and results are showing up!!!

Muse is just awesome🔥

Meta cooking up, while @Google still not releasing Argon to public!!!

[Mark Zuckerberg @finkd]

Introducing Muse Wonder, our new family of frontier models with state of the art performance for personal and professional agentic work.

[Sudarshan Shah @sudarshandev119]

@OpenAI & other labs should better have consensus with international community of mathematicians, on how Math research to be conducted, how things work, rather than dumping proofs, which is not gonna do any good to the Math field or anyone.

[Sudarshan Shah @sudarshandev119]

This is good. Math research is not just about getting end result, it's also about process of research, every line of proof matters, why it is there matters more.

Everyone thinking it's like Software Engineering, where end product is important & code can just be written by AI.

[internet hall of fame @InternetH0F]

OpenAI pulled its $1 million sponsorship of an AI math contest at Caltech after mathematicians announced a boycott

An open letter signed by 771 mathematicians called the event "slop mathematics," as they warned AI labs are harming the field

[Sudarshan Shah @sudarshandev119]

@piyushgarg_dev @rishikagupta__ Congratulations @piyushgarg_dev and @rishikagupta__ .
Best wishes for the future👍

[Sudarshan Shah @sudarshandev119]

This is absolutely crazy good release, as now we can spawn subagents with Opus 5.5 as main model.

@AnthropicAI has right now best lineup of 5.5 family models release, which are capable, cheaper & really good in actual task with nice daily & weekly limits.

Good work @claudeai

[Claude @claudeai]

Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released.

On average, it costs around 75% less to run than Claude Haiku 4.5.