Aiterator | SDE | 4x Hackathon Winner | NITK '25
Bengaluru, India
Joined May 2023
- Tweets192
- Following256
- Followers66
- Likes472
Pinned Tweet
ETHGlobal Finalist in the City of Dreams!
We conjured Air Agents - A novel way to interact with your agents in the spatial and audio domain.
- top 8 overall
- @hedera SDK track winner
- @worldnetwork 2nd place
Couldn't have done it without @ahmedfahim21_ & @AbhishekSa6742🙌
Thanks to @ETHGlobal for this amazing opportunity♥️
I wonder if there are more "planets" out there, in our solar system, that we haven't seen.
smithsonianmag.com/smart-new…
2
The history of taste in research through time, is the new place to look for more creativity. I agree that variety and exploration are the key to progress.
resobscura.substack.com/p/re…
2
68
A mind-blowing way to fingerprint GPUs 🤯
I never through mathematics and optimization would be used to track supply chains.
unemploymentlabs.com/checkpo…
1
5
69
Eventually LLMs (or SLMs) will run on wearables. Probably for health related data, and other privacy requisite data.
github.com/hyperionaidev/smo…
6
197
I wasn't able to try our #Fable 5.2 over the weekend, since we were busy hacking. Now, I used it to revamp my portfolio website:
chinmayasahu.com/
It works like a charm, and tells my Fable :)
2
12
406
Chinmaya Sahu retweeted
🇮🇳 ETHGlobal Mumbai Finalist | 🫨 Air Agents
A new way to interact with your agents in the spatial and audio domain.
Built by @AbhishekSa6742, @ahmedfahim21_, @csking101
🔗 air-agents.vercel.app
🏆 ethglobal.com/showcase/air-a…
Happy to be a part of this @ETHGlobal , it was a pleasure hacking along with @ahmedfahim21_ @AbhishekSa6742 and building Air Agents🫨
Dropped in at the @noderails booth during @ETHGlobal Pragma!
Payment infrastructure making crypto more practical for everyday use: gasless checkout, stablecoin settlements and chargebacks.
Enjoyed learning about @wallcardapp too!
@chiefmohitbhat #NRL #NodeRails #Pragma
LLM-CRISPR 🤯🤯🤯
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math.
rikiyat.github.io/blog/weight-sp…
github.com/RikiyaT/lm_und…
huggingface.co/Qwen/Qwen3.5-0…
huggingface.co/RikiyaT/Qwen3.…
4
135
Chinmaya Sahu retweeted
built something over the weekend with @openuidev & @SarvamAI
pause any NPTEL lecture and ask your doubt. the model generates interactive answer UI live with key ideas, flashcards, quiz, timechips, etc. in your language
Chinmaya Sahu retweeted
Finally releasing the results of the arch ablations I was running with bux over the weekend
DOOMformer, and example of how NNs are just a composition of functions, which you can hand craft.
github.com/mottopanikeiku/do…
65
Is an update, always an upgrade 🤔
fortune.com/2026/10/30/cisco…
61