Aditya Goyal ✈️ ICML'26

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@prof_kamilov As an undergrad , should I include my extensive swe experience . I am applying for masters research

[Ankit Gupta @agupta]

Live in the future. Build what’s missing. @ycombinator

[shyamal @shyamalanadkat]

after close to four years at @openai, i moved from the bay area to india earlier this year. i still believe deeply in ensuring true superintelligence accelerates science and remains accessible and beneficial to all. having grown up here, i've also always felt deeply connected to the ecosystem here.

over the past several weeks, i've been speaking with researchers, engineers, and thinkers across india and apac. it's become clear that there are many who want to build the future from here. moving back felt like the counterintuitive choice. i no longer think that's true.

what's been missing is the belief that you can build institutions of global consequence from anywhere. and more importantly, the ambition and the will to pursue ideas that seem impossibly large at first. this may be a once in a generation opportunity.

more to come soon. DMs open if this resonates.

[himanshu @himanshustwts]

It was great evening hosting the conversation on the future of science communication w/
- @tensorqt (paradigma),
- @rehaanahmad171 @rajpalleti314 (alphaXiv) and
- @bohannon_bot (science journalist)

notes from the convo:

+ there is a risk of “vibe science” era where science loses its intrinsic passion and meaning. more about a future where machines dominate certain sciences (like theoretical physics) leaving fewer humans able to judge or appreciate breakthroughs

+ scientific papers traditionally serve multiple roles designed for scarce human attention but now research volume has exploded beyond expert capacity.

+ readership patterns have changed now. short attention spans on research papers with most papers engagement plateauing after 12 weeks except for a few that gain traction longer

+ human readers vs agent readers pose different design challenges with current systems focused on human usability but evolving toward automated agents for literature review + decision-making

+ science is bounded by bandwidth and scale with hypotheses needing real-world prediction and testing. research goals and results are socially negotiated (not absolute) reflecting varied researcher perspectives.

+ current models struggle with reward hacking and reproducibility (which brings necessary human checks). the future may involve agents building consensus and ratifying results in a global verification tree. automated peer review is a long-term goal but requires rigorous evaluation to avoid errors.

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@slowdownisha @lossfunk Great session indeed !

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@DrEmmaZang GOAT GOAT advice. Be genuinely curious in what you can learn from the other person.

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@nitishfy A reason why innovation is not in products even for which India has infra is because people actually believe they need an IIT degree to do a startup lol . It’s a fricking mental block that needs to be removed .

[Nathan Lambert @natolambert]

My book, Reinforcement Learning from Human Feedback is done!

This is the book I wish I had when learning to fine-tune, align, & now post-train models since ChatGPT. The resource has been built by me finding time to study and document the fundamentals on nights and weekends since 2024.

Transferring as much of the intuitions of building Olmo as I possibly can in the book format.

The book is launching with an over 10 hour, full course with slidedecks, functional code for the training chapters, an example model completions library, and of course the free online web version.

Physical orders from Manning will ship in 1-2 weeks, and Amazon a week or so after. Thanks for your support!

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@siddarthpaim we have the talent and money to build the companies on these models. we are seriously lacking on the applications side as well.

[Indian Tech & Infra @IndianTechGuide]

🚨 Successful launch of India’s 1st private company rocket, Skyroot Aerospace. 👏🏼🇮🇳

[Avi Trost @atrost3122]

Find me at poster 1713!

[Tongtian Zhu @Tongtian_Zhu]

ICML 2026 has been the most rewarding conference experience of my PhD so far.

Social events and dinners were so packed that I was constantly running from one to another, surviving on four hours of sleep per day. But it was absolutely worth it!

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

Had my very first poster session in my research journey at ICML yesterday. Inspired and grateful ! Some people from Google deep mind stopped by 😁 and they had some hard hitting questions that I enjoyed answering. Gonna be one at the main conference for the next ICML 💪.

[Tzu-Heng (Brian) Huang @zihengh1]

Have a great time at the WSS workshop today. Thanks for organizing it—love the lineup!

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@jfrcooper2 Congrats !!

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@arimorcos @datologyai Hey Ari . Would love to chat! I am really interested in the work that datology ai is doing and would like to know more about how I can get into data curation

[Ari Morcos @arimorcos]

Just landed in Seoul for #ICML2026!

Please reach out if you're interested in chatting about all things data, from filtering to scaling to synthetic, or if you'd like to learn more about working with us @datologyai to understand the fundamental principles of data curation!

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@iarthsingh Thanks buddy. We should meet in Korea and chat over a cup of coffee

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

A Google deep mind researcher once said , researchers can read other papers , try to combine methods and produce something with better results on some controlled experiments or baselines and publish , only to end up having their idea not being implemented. I think thinking more about ideas that do not build off other ideas in a very fine granular way and experimenting the idea on real world data instead of only controlled experiments is the way to go. But a lot of researchers don’t follow the value framework as it doesn’t really lead to guaranteed value. So i think practically , a combination of how confident you feel with the idea and a good balance of optimizing for value and for papers is the way to go. Which so difficult to do and must be internally internalized.

[Aditya Goyal ✈️ ICML'26 @adigoyal0807]

@kimbochen toy story's fire but our conversation is gonna be more fire. lets chat ! i got some cool ideas to discuss.