Aditya Timmaraju
@tadityasrinivas
Sr Staff MLE @GoogleDeepMind on Gemini efficiency. Ex @meta, Grad @stanford, @IITHyderabad. 2013-24 in SF Bay. Karma Yoga. Definite optimism.
Great to see the power of Matryoshka🪆 architectures, devised by my lab, getting widely highlighted as an indispensable component of the LLM cognitive core. cc @jainprateek_ @adityakusupati A sample list of Matryoshka papers from my lab: Matryoshka Representation Learning…
The race for LLM "cognitive core" - a few billion param model that maximally sacrifices encyclopedic knowledge for capability. It lives always-on and by default on every computer as the kernel of LLM personal computing. Its features are slowly crystalizing: - Natively multimodal…
Great to see the amazing @SarvamAI team building upon the wonderful work from our team, led by @partha_p_t, and our collaborators to build really powerful models for Indian languages.
Great to see improved Indic language tokenization & processing capabilities in Gemini & Gemma models helping the ecosystem build further. Congratulations to @SarvamAI on the Sarvam-translate launch! #AIForAll
Drastic progress on maths with Gemini 2.5! As a math undergrad, I am impressed 🤯 🥈 -> 🥇 ✅ Formal -> Informal ✅ Specialized model -> General model ✅ Available soon ✅ Huge thanks to IMO and congrats to all participants! Blog: deepmind.google/discover/blog/…
Happy to see wonderful contributions from our team, making Gemini 2.5 models more powerful, more efficient, and understand many more languages and cultures!
It's extremely gratifying to see so many contributors to Gemini 2.5 from @GoogleDeepMind India! Sip filter coffee (w/ plant-based milk, of course) as you pave the path to AGI, can't think of a better deal 🙂 (Btw, we are growing, app link coming soon!) arxiv.org/abs/2507.06261
8. Prediction: In the next decade, an increasing number of mathematicians will improve their productivity by using LLMs to search for known parts of a tentative proof outline. The old guard mathematicians may lament this, but the young folks will just go on to do good work.(8/10)
Great to see Gemini ✨come out on top! This is a verifiably uncontaminated benchmark (meaning not exposed during training) comparing models on the latest International Math Olympiad (2025) 🧮. We’ll find out tomorrow if this score fetches a Bronze because human candidates’ test…
We just released the evaluation of LLMs on the 2025 IMO on MathArena! Gemini scores best, but is still unlikely to achieve the bronze medal with its 31% score (13/42). 🧵(1/4)
From "After the Spike": "The year 2012 may well turn out to be the year in which the most humans were ever born—ever as in ever for as long as humanity exists." Going to live-tweet interesting bits.
Given the switches have brackets and a stop-lock mechanism, it is virtually impossible for a human to turn both the fuel switches to “cut-off” mode within a 1 second timeframe of each other (which we know from black box readings). Both Reuters and BBC’s coverage completely…
Hey @BBCNews @BBCWorld this is absolutely incorrect - nowhere in the Air india crash report does it say the pilots cut off fuel to the engines and that there is no fault in the plane. Pretty shocking to see this scurrilous headline. In fact the report mentions an FAA advisory on…
Very excited to share that @windsurf_ai co-founders @_mohansolo & Douglas Chen, and some of their talented team have joined @GoogleDeepMind to help advance our work in agentic coding in Gemini. Welcome to our new team mates from Windsurf! theverge.com/openai/705999/…
Excited to see AI Mode go live in India!
After an incredible response in Labs, we’re starting to roll out AI Mode in Search to everyone in India (English to start). It’s a total reimagining of Search, and we’re excited for even more people to use it.
Having been a fan of fireworks during each Diwali and New Year’s growing up, I’d always look forward to the 4th of July fireworks shows during my life in the USA. Two of the shows that I saw stand out. One was at Lake Tahoe, renowned nationwide for its fireworks show, for my…


Quotes that didn't age well: "Playing him in this tournament I will approach it as if I'm playing one of the presumably weaker players" (Magnus on Gukesh 2 days ago). LOL.
Feedback of this kind should be taken seriously and acted upon.
I have to file more useless paperwork for our entity in India than all our other countries combined.
2025 founder flexes
Putting the application email from Soham directly in my investor deck. They're agog
Nearly a $12B fund for incentivizing R&D in deep tech startups. At a quick glance, looks like both debt and equity based financing are within scope. Hopefully, the implementation is smooth and leads to a deep tech startup spring in India. 🇮🇳 🚀
This is the biggest announcement of the year. #RDI 🇮🇳
We are hiring Technical Program Manager to organize and enable our research teams to be the best at what they do and to make fast-paced progress towards our mission of bringing AGI responsibly. Ideal candidates should have a demonstrable record of strong program management…
With the Gemma 3n model being open-sourced, look forward to developers on @Android and in other ecosystems unlocking the full access potential that the Matformer architecture enables!
📢Now open, Gemma 3n weights & it is natively flexible, first of its kind, thanks to MatFormer🪆 Any model between E4B & E2B with ZERO training near Pareto -- we found a bunch! Find a better E3B than what we released, I will send you a 🪆😉 Find the colab for extraction 🧵👇🪆
Hard evidences extending the length of our civilization just keep getting ‘unearthed’
In case anyone hasn’t grasped how much of a big deal this is, the depth excavated is 23m. By comparison, Göbekli Tepe is 15m for 9600BCE. This is why the article mentions Sarasvati settlement since the Holocene era.
AI Mode is coming to India – our first international expansion! Labs users in India can now start testing our most powerful AI search, designed to help with your toughest questions. Exciting times ahead for Search … stay tuned for more this summer.
“The ball disappears behind his elbow and the batter momentarily loses sight of the ball…It's like a catapult. At the very last millisecond, the ball is back in the batter's view, hurtling at somewhere around 90mph.” bbc.com/sport/cricket/…