Aditya Timmaraju
@tadityasrinivas
Followers
3K
Following
80K
Media
26
Statuses
627
Sr Staff Research Engineer @GoogleDeepMind Frontier AI. Ex @meta, Grad @stanford, @IITHyderabad. 2013-24 in SF Bay. Karma Yoga. Definite optimism. Erdős no 3.
Hyderabad, India
Joined May 2010
Is the 'reverse brain drain' in tech gaining momentum in the wake of visa upheavals? I shared my own story with @moneycontrolcom: after over a decade in Silicon Valley of working on cutting-edge AI, I moved back to India last year. Inspired by PM @narendramodi's call to NRIs to
moneycontrol.com
Rising nativism in the West has taken the sheen off building a life there. An Indian techie who decided to come back writes that there’s never been a better time to get the best of both worlds: the...
OPINION | 'Making My Janmabhoomi My Karmabhoomi': Why this tech NRI came home @tadityasrinivas ✍🏽 https://t.co/dYSiHp3iia
1
3
34
@tadityasrinivas Okay, I agree that my initial wording was somewhat tendentious. Perhaps see @ysu_nlp’s post for a more calm and balanced take. 😊 But its roots lie in lots of recent posts, podcasts, videos, etc. arguing that language models are really limited and how much we need so-called world
psychiatrist.com
A review of more than 2,700 trials confirms that regular exercise boosts cognitive function, especially memory and executive skills.
1
6
29
Natural language is an encoding/expression of meaning and logic of experiences that are central to humans. So it’s not surprising in hindsight that language modeling leads to emergent capabilities in tasks that *we* consider important. I don’t see the need to compare the
Math/physics-envious crowd still trying to cope with how central languages—human and symbolic codes—are to modern human existence. Occasionally I try to toss a wrapper into a garbage bin but usually I’m discussing, writing plans, coding, thinking through a strategy, ….
0
0
0
Good thing there’s been a significant drop in the drain of India’s engineering and STEM graduates post-pandemic. Tightening immigration just further cemented this trend. Long India. 🇮🇳
What if we split Asians by country of origin? We find that Indians are the most economically positive major country of origin group. Chinese immigrants contribute similarly to other Asian immigrants but Indians rise above the pack.
0
0
10
Someone say “intelligence too cheap to meter”? ⚡️
Gemini 3 Flash on the @ArtificialAnlys intelligence benchmark, the most cost per intelligence efficient model in the world!!!
0
0
2
We’ve pushed out the Pareto frontier of efficiency vs. intelligence again. With Gemini 3 Flash ⚡️, we are seeing reasoning capabilities previously reserved for our largest models, now running at Flash-level latency. This opens up entirely new categories of near real-time
50
195
2K
I’m flattered (?) @OpenAI, except I haven’t really applied 🤔😄 1) someone impersonated me and used my public CV 2) there’s a bug in their system that randomly samples DeepMind folks and sends these out 3) a new "Jedi mind trick" recruiting strategy to coax people into the
21
4
262
Nice to see this support for AI research and innovation in India.
It was a privilege to join Hon'ble Minister @dpradhanbjp Ji, Sh. Vineet Joshi, Preeti Lobana, and leading researchers & innovators to announce our support (including grants worth $10M) to the AI CoEs and the Pushpak Bhattacharyya Centre on Indic Language Technologies at IITB.
0
0
9
Got to witness my favorite footballer from ever since I started watching football 17 years ago and my fav club @FCBarcelona and @Argentina legend, @TeamMessi’s magic with the football from 20 meters away, with @LuisSuarez9, another Barça title winner. A bucket list item checked
1
0
14
🚨 HIRING ALERT! 👩💻👨💻 We are hiring for both Pre-doctoral and Post-doctoral AI research positions at @GoogleDeepMind India🇮🇳! This is a great opportunity to work on fundamental research problems in AI and impactful applications in @GeminiApp and beyond. Our past predocs and
arxiv.org
Quantizing model weights is critical for reducing the communication and inference costs of large models. However, quantizing models -- especially to low precisions like int4 or int2 -- requires a...
14
32
424
Created one for West Hyderabad following this viral trend with #NanoBananaPro 🍌 @GeminiApp
Nice Nano Banana Pro prompt for weather app: CITY=Prague,Czechia Present a clear, 45° top-down isometric miniature 3D cartoon scene of [CITY], featuring its most iconic landmarks and architectural elements. Use soft, refined textures with realistic PBR materials and gentle,
2
0
9
TPUv7: Google Takes a Swing at the King Potential end of the CUDA Moat? Anthropic’s 1GW+ TPU Purchase, The more (TPU) Meta/SSI/xAI/OAI/Anthro buy the more (GPU capex) you save, Next Generation TPUv8AX and TPUv8X versus Vera Rubin, All the 3D Torus Cube Diagrams you could
43
118
800
+ business leaders
Holy shit. I’ve used ChatGPT every day for 3 years. Just spent 2 hours on Gemini 3. I’m not going back. The leap is insane — reasoning, speed, images, video… everything is sharper and faster. It feels like the world just changed, again. ❤️ 🤖
0
0
0
Med guys
Eureka moments in generative A.I. Nov 30, 2022: ChatGPT Nov 22, 2025: Gemini-3
0
0
0
More echoes of a GPT-4 like leap with Gemini 3 from “high taste testers”, to borrow someone’s phrasing 👀
the Gemini 3 moment seems closer to GPT-4 than any other moment in recent times. So much progress released so suddenly that it's overwhelming. Especially NanoBanana's progress. OpenAI around GPT-4 looked invulnerable , and Google looks invulnerable now with Gemini 3 (+TPUs
2
0
3
Tell me this isn’t insane
You went 🍌🍌 for Nano Banana. Now, meet Nano Banana Pro. It’s SOTA for image generation + editing with more advanced world knowledge, text rendering, precision + controls. Built on Gemini 3, it’s really good at complex infographics - much like how engineers see the world:)
179
681
10K
Depicting a fascinating real-life tale of one of my favorite GOATs Ramanujan in a comic strip, thanks to Nano Banana Pro.
0
1
24
I was deciding between offers from @Google and Microsoft when moving back to India last Jan. It was still during the “Bard” phase. And Microsoft was flying high with OpenAI. With due respect to MSFT AI folks, my choice was to bet on @GoogleDeepMind’s deeper talent pool. Nearly
There’s a new sheriff in town. Google just passed Microsoft in market cap.
16
9
918
Yup, it’s not even close. Go try Nano Banana Pro 🍌 for all your fun and infographic needs.
Gemini 3 Pro Image only has an 8 percent error rate when generating text OpenAI's model is at 38%
0
0
3
Banger
btw, you can bring your graph back to reality. You are welcome.
0
0
5
Now that the cat is out of the bag, made with Veo
@OriolVinyalsML spilled our super-secret recipe for making Gemini 3 better than Gemini 2.5: improve pre-training and improve post-training! 😅
0
0
2