@devvarma

Professor, Former Dean R&D @iiit_hyderabad Enthusiast: Human and Machine Learning and Knowledge Ecosystems

 Hyderabad, India
iiit.ac.in/~vv
Joined January 2009
  • Tweets1.9K
  • Following183
  • Followers2K
162Media
Include

Only show posts containing:

Exclude

Hide posts containing:

Time range
-
Minimum likes
Shreesh was my lab mate in IIT Kapur when he was doing his PhD - many of us admired his intellect and benefitted from his generosity. It is so good to hear about him a long time. moneycontrol.com/education/i…

IITian who became a Monk: Meet Shreesh Jadhav — engineer who gave up a million-dollar future to...

Shreesh Jadhav’s inspiring journey from IIT Kanpur topper and GATE ranker to monk and educator. Explore how he redefined success through service and purpose.

moneycontrol.com
9
612
Vasudeva Varma retweeted
IIIT Hyderabad@iiit_hyderabadFeb 28
IIIT-H alumnus Siva Reddy @sivareddyg , now a professor at @Mila_Quebec, is shaping the future of NLP. A 2026 Sloan Research Fellow, he shares his journey and what drives him. blogs.iiit.ac.in/siva-reddy/
2
1
18
937
The Panini Linguistics Olympiad (PLO) is open to students of Grades 6–12. 🔗 ltrc.iiit.ac.in/plo/
3
1
353
"...old Silicon Valley adage-if you aren’t paying you are the product. In AI, we’ve been paying steeply for the product, but we still are the product." openfuture.tenstorrent.com Powerful blog from @tenstorrent : wakeup call to build DPGs and Opensource platforms.

Open Future

We are building an open future for AI. Own your silicon future. Join us.

openfuture.tenstorrent.com
110
Replying to @devvarma
@devvarma opines that India must build cost-competitive, resilient cloud alternatives, but stresses AI is inherently global. He advocates sovereign cloud development while staying engaged with the worldwide AI ecosystem. tinyurl.com/mtra48ct
1
2
110
Vasudeva Varma retweeted
Deedy
@deedydas31 Aug 2025
This new DeepMind research shows just how broken vector search is. Turns out some docs in your index are theoretically incapable of being retrieved by vector search, given a certain dimension count of the embedding. Plain old BM25 from 1994 outperforms it on recall. 1/4
87
374
46
4,101
475,948
Load more