Smallest AI is in Vegas for Reinvent!

Text-Speech(Character-Frame) alignment is the hardest problem I have ever worked on.

Meet the Smallest Gang at AWS re:Invent 2025, at Booth 177. Get to know how Voice AI is transforming Enterprise operations. P. S. - We are hiring speech researchers, and have referral rewards too. Feel free to drop for by and refer your best buddies. Sudarshan Kamath Apoorv Sood

Also at #NeurIPS2025 , we’re hiring Speech Researchers at smallest.ai. Happy to chat if you’d like to learn more.

Learn how to create robust production agents! Rounak & Aditya Bhat go in-depth into: 1. Creating a Credit Card Inbound Agent 2. How to prompt the agent 3. Managing edge-cases DM either of them if you'd like to get the prompt used in this video 🚀

We are at Neurips, let's talk Speech, memory and AGI 🥂 Wasim Madha @ NeurIPS 2025 Hamees @ NeurIPS smallest.ai Pro tip - always register a day earlier 🤣

Never thought in my life I'd be at a NeurIPS 2025 in San Diego, and here I am. Attending my First Talk at #neurips25: Foundations of Tensor/Low Rank Computations

We're going to be at NeurIPS Conference 2025. And we're excited to meet some cracked folks! Hit up Akshat Mandloi Wasim Madha @ NeurIPS 2025 Hamees @ NeurIPS who are attending the event. We are a also recruiting for the following roles: 1. Data Scientist (Voice) 2. GPU Optimization Engineer

First time at NeurIPS and I get to present my paper. Feels crazy to go from reading this conference from afar to actually being here. Grateful, excited, and ready to learn from everyone around.

Oh we love your merch”! “This bag is so cool!” “I love your branding!” “Is that really AI?” Sentences heard today at AWS Events re:Invent in our 90 min speed round of demos and merch But all this takes an army. Some special folks who made it happen are tagged below :)

The general norm has become to use a very large model and show it enough data to let it discover the patterns itself. Well, applying a few heuristics and basics of a domain can lead to significant reduction in parameters and data required. Watch Sudarshan Kamath describe this in

Looking for good design engineers. Want to refer a few to teams and hire one myself. Drop portfolios below or tag people you like.

"Linear Algebra for Machine Learning: From Scratch to Mastery" series: there will be 40 lectures of around 1 hour each for linear algebra, syllabus in the comments. Series Philosophy: This lecture series is designed to be the definitive, one-stop resource for students,

Shipping something absolutely banger Devanshpawan

Everyone is chasing bigger models. We’re betting on precision. Here’s what Sudarshan Kamath (Founder & CEO, smallest.ai) had to say about where AI space is really heading in the next few years. Watch the video below ⬇️

STREAMING LIVE, RIGHT NOW! 👇 announcing the world’s first LLM mafia livestream, inspired by ClaudePlaysPokemon. twitch can chat live with the LLMs as they lie, manipulate, and make deductions. watch Gemini, Claude Opus 4.5, GPT 5.1, and 7 more models battle it out. Built with

Wrapping up NeurIPS with my first poster session and it went amazingly well. Met brilliant researchers, mentors and had some great conversations. Grateful for the experience, looking forward to many more :)