Anuj Dube

SRE/MLOps
| Infra, GPUs

[Anuj Dube @anujsdtw]

After 15 months of building…

My first SaaS Uptimebeats got acquired 🎉

• First dollar on the internet
• ~$600 revenue
• Sold for $4,750

Small exit. But Learned lots of big lessons.

On to the next 🚀

[Anuj Dube @anujsdtw]

I need to actually start learning deeply on llm inference. Big skills issues on that part.
Just too many things to understand

[Anuj Dube @anujsdtw]

Customer support is very hard to get good considering this billion dollar companies are also failing at it.

[Anuj Dube @anujsdtw]

With my limited experience with enterprise support

Nvidia support is understandable,
AWS support is ok,
azure is bad and
gcp is just shitty.

Maybe there is reason AWS is market leader.

[Anuj Dube @anujsdtw]

By mistake opened dhan us stocks account. Wanted to explore app and it just opened new account with my already stored kyc. Now one more account to handle compliance 😬

[Anuj Dube @anujsdtw]

@VazeKshitij On bright side it's a construction debris and doesn't smell 😅. It's just infront of hdfc bank office.

[Anuj Dube @anujsdtw]

@SinghDevHub Isn't that one of the measure they use for preventing exam getting leaked.

[Anuj Dube @anujsdtw]

@SinghDevHub Reason might be cheating but also fairness. Creating rule and not applying it to everyone is just unfair to people who follow rules. Free education/hospital are different topic but just too much expectations from 3k gdp per capita country.

[Anuj Dube @anujsdtw]

@zarazhangrui I might just bookmark it again for reading it next time

[Anuj Dube @anujsdtw]

@vijayshekhar Atleast 60 on artificial intelligence and cheaper than current model on same level launched before next 3 months, so it's almost impossible.

[Anuj Dube @anujsdtw]

Curd and yogurt are different 😮
Until now I used to think it's just same stuff with different name

[Anuj Dube @anujsdtw]

@cneuralnetwork 2nd one
Adult human brain weights around 1.3kg

[Anuj Dube @anujsdtw]

@levelsio Hope this was true.

[Anuj Dube @anujsdtw]

@vansh22b @varmology Last financial year their revenue was $5.4m

[Anuj Dube @anujsdtw]

Why it's under non-commercial license
What's even point of these models?
Who will even adopt these models even if they wanted to.

[IIT Bombay @iitbombay]

AI that thinks in India's own languages.

IIT Bombay is proud to present BharatGen to the world: Open, multilingual AI for India's languages and people, at Bharat Innovates 2026 in Nice, France (14–16 June).

BharatGen is built at IIT Bombay's Department of Computer Science and Engineering, led by Prof. Ganesh Ramakrishnan, with Rishi Bal (CEO) and Dr. Maneesh Singh (VP, ML) with a consortium of 9 premier academic institutions. A team of 60+ researchers, engineers and linguists are building AI that includes all scheduled Indian languages, across text, speech and documents.

-> Param2, its foundational text model with reasoning, coding, and tool calling capabilities works across all 22 scheduled Indian languages

-> Shrutam2, for automatic multilingual speech recognition/ STT across Indian languages

-> Sooktam2, a text-to-speech models with zero-shot voice cloning across Indian languages

-> Patram, a document vision model built for understanding Indian-specific documentation

BharatGen powers services in governance, healthcare, education, insurance, finance, and cultural preservation.

A national effort backed by DST and the IndiaAI Mission, BharatGen is India's push for open, homegrown AI, built for 1.4 billion people.

For more information, visit https://t.co/bZul5Lr3yC

Bharat Innovates 2026 · 14 - 16 June · Nice, France

@BharatInnov2026 @EduMinOfIndia

#BharatInnovates2026 #IITBombay #BharatGen #DeepTech

[Anuj Dube @anujsdtw]

@GutsB35378 @HarveenChadha Government doesn't own any GPUs. They are giving money to companies like yotta who gives GPUs to companies like sarvam for cheaper than market. It's just subsidy. Government is also taking some equity for this so not for free.

[Anuj Dube @anujsdtw]

@AdithyaGMenon @KrishnaPra60146 Government repo shows 13% in 2016
https://t.co/speTtlRT4x

[Anuj Dube @anujsdtw]

@GutsB35378 @HarveenChadha highly likely capacity is already booked. As most investment is from meta, gcp,azure,AWS. Sarvam does rents servers from yotta last run was done with their help. Yotta will have 80k GPUs around 2027 so they will rent more from yotta.

[Anuj Dube @anujsdtw]

@antirez gpt-oss is still one of my favourite model. Waiting for next open source model from OPEN AI.

[Anuj Dube @anujsdtw]

@GutsB35378 @HarveenChadha 100k H100 equivalent data cluster will cost ~$10B. they can train in 2–3 months & iterate few times. They can do it with $1B cluster and a longer timeframe also. But the bigger problems will be data sets, research, and buying GPUs. xAI has 200k gpu but you know the results.