Ajay Rayasam (@ajayoup) header
Ajay Rayasam (@ajayoup) profile picture

Investor @oup_vc, @MIT alum (grad 2020)

San Francisco, CA
Joined February 2022
Congrats to @exnx, @MichaelPoli6, and the rest of the @RadicalNumerics team!
Introducing RND1, the most powerful base diffusion language model (DLM) to date. RND1 (Radical Numerics Diffusion) is an experimental DLM with 30B params (3B active) with a sparse MoE architecture. We are making it open source, releasing weights, training details, and code to catalyze further research on DLM inference and post-training. We are researchers and engineers (DeepMind, Meta, Liquid, Stanford) building the engine for recursive self-improvement (RSI) — and using it to accelerate our own work. Our goal is to let AI design AI. We are hiring.
1
1
9
747
Ajay Rayasam retweeted
We’re excited to introduce RunLLM v2 today! 🎉 RunLLM v2 is rebuild of the product from the ground up focused on delivering the most powerful and flexible platform for enterprise support teams. Read the RunLLM v2 launch blog post: 👉 tinyurl.com/49st5ns2 Today’s launch includes: 🤖 A new agentic planner with fine-grained reasoning and tool use support ✨ A redesigned new UI that enables creating, managing, and inspecting multiple agents ⚙️ A Python SDK that allows you to exercise fine-grained control over support workflows We’ll be sharing more throughout the week, but today we’re focused on how RunLLM’s new agentic capabilities enable more precise answers and more effective debugging.
4
14
7
33
9,651
Congrats to Snowcap Compute on announcing their $23M seed round led by @Playground_VC with participation from @OUP_VC, Cambium Capital, Vsquared Ventures and others. We're hitting physical limits of conventional silicon and Snowcap is attempting to breakthrough with a novel approach — superconducting chips that can run classical algorithms when cryogenically cooled. I love backing teams who've spent years, or even decades, obsessed with a contrarian idea. They shut out the noise and quietly build until their moment arrives. Long ways to go, but excited for what’s next. Privileged to work with the team and @PGelsinger on his first deal! reuters.com/business/snowcap…
1
10
45,270
Ajay Rayasam retweeted
Founders who were PhD or post-doc in my lab at Berkeley, **largely funded by NSF / DoD grants**, start-up, market cap (collected by OpenAI Deep Research)
118
473
120
4,721
1,015,784
To all aspiring DeSci founders and daos out there - send that cold email. Tell your scientific idols about DeSci, and not why it helps them raise funds but what new collaboration models it enables If you want help shoot me a DM of your email and I can give feedback 👩‍🔬 best submission gets 1000 $BIO
Anyone who says DeSci is a scam is jaded by short termism and has lost the values that created Bitcoin and Ethereum. Solving long term problems for real people. @bioprotocol first BioDAO on Solana is by Harvard alumni and biotech legend @eperlste who has dedicated his life to helping families find cures for rare diseases. Auction is reaching 10,000 SOL and closes in the next hour. I want to thank the @solana community for the warm welcome during an incredibly tough downturn. Its proof people want utility and believe in DeSci. Build long term @cz_binance cautioned recently. And here's the story: I first emailed @eperlste in 2019. Ethan was an inspiration to me early on when I looked for answers to a crisis I had seen since my teenage years. At 14 I asked myself why family members were prescribed anti depressants or adderall - I felt this was a failing system. I fell into a rabbit hole of online biohacking forums and geeked into all sorts of things from longevity to nootropics and genetics. I first cold emailed him in 2019 when I founded @Molecule_dao after going to the only community I could discover about what I wanted to build. His talk was so inspiring I wanted to learn from him. I emailed him twice and second time told me he didn't have time to give advice. The Open Source Pharma foundation launched by a policy prof at Harvard with an Indian institute simply because they decided they wanted to develop drugs without patents and work in github repos. They accelerated numerous drug candidates in tropical diseases and malaria. Gates foundation funded a stage 1 trial with very promising results. But nobody wanted to ever fund a stage 2 trial because there was no potential commercial strategy. We tried to launch a DAO with them, but The basic premise makes so much sense - if we did drug research in the open, scientists collaborate and we accelerate cures. But when I heard that a promising malaria vaccine funded by the gates foundation had no hopes of ever making it to trials because nobody funds drugs without patents I realised how broken the system was. Anyone who tells you biotech research should be open doesn't understand what the system incentivises for. It incentivises for returns, not impact or outputs. Not cures. It punishes open innovation because this threatens the margins. But Ethan tried it with @PerlaraPBC which you can find many stories about. He pitched biotech VCs a new model of a public benefit corporation incentivising data sharing among families. Unfortunately, post seed stage biotechs need to play by the rules and I recall he failed to raise follow on funding. But he learned that patients are champions, they make it happen. In 2021 Ethan emailed me out of the blue and proposed I connect with another rare diesease founder excited by VitaDAO. @VibeBio was focusing on this use case and we had enough to do with our first dao so things went dormant again. Brilliantly @EMcCarthyPage invited Ethan to speak at DeSci Berlin last year and he gave the best talk in my view and laid out the vision for the DAO. Its simple: 1. Rare diseases are not lucrative for industry bc populations are so small, but its the life and death of a child and could be prevented. No insurance will cover you, so you're basically an entrepreneur in this disease now. @blader recently became on $MIRA turned into a beautiful story. 2. Rare diseases can be cured but scientists need data and impulses from patients and a bit of funding. What normally happens in board rooms and strategy meetings and expensive labs is now run by a father trying to save his daughter at every cost, and prevent anyone else from facing this. 3. Ethan realised you can cut the costs down massively by helping the parents project manage, managing the data safely which is sensitive, and budgeting costs and refining drug development. Sharing data is sensitive and why trust crypto, patients have made bad experiences with data sharing powered by $CURES Ethan got a therapeutic into stage 3 FDA for under 5 million. That normally costs billions and takes years. Regulator actually works with you here and fastracks because the need is so big. @bioprotocol gave Ethan and team builder grant in Q3 last year so he could accelerate his work. We knew we wanted to launch. Instead of building a dao, he went out took the 100k and said let me show you what this gets you. He ran a yeast screen and discovered 2 novel targets and filed two patents. When you ask, does $CURES have a product its selling? Bitcoin doesn't sell a product. It provides an alternative to a massive real world problem. @endrarediseases provides a system that could generate 1000s of products for millions and would've never been previously possible. And the consumer will hold $CURES for life because they care about access not profit. Incredibly grateful to @akshaybd @solana @rajgokal @nkumar23 for encouraging us to bring DeSci to Solana! Onwards
15
6
82
8,294
Ajay Rayasam retweeted
Nice
127
31
8
420
23,978
Ajay Rayasam retweeted
We've raised a $64M Series A led by @kleinerperkins to build the platform for real-time voice AI. We'll use this funding to expand our team, and to build the next generation of models, infrastructure, and products for voice, starting with Sonic 2.0, available today. Link below to try it free 👇
110
105
69
1,058
329,431
Congrats to my friends @vsreekanti, @profjoeyg, @joe_hellerstein , @cgwu0530 and the rest of the @RunLLM team on their public beta launch. Check it out!
The RunLLM Public Beta is Live! 🚀 After almost two years of work, we’re launching RunLLM — the first AI Support Engineer. Built for advanced technical support, RunLLM: ✅ Saves support engineering time ✅ Accelerates customer adoption ✅ Generates customer and product insights We’ve also designed an awesome new onboarding experience that we can’t wait for you to try — go to the RunLLM website to sign up and try it for free.
3
4
476
Ajay Rayasam retweeted
Honored to be @lexfridman podcast with the homie @natolambert 5 hour podcast dropped and it's a banger covering a lot of different topics. We had so much fun riffing. The secret to the hostage images Lex takes are now revealed with this nato selfie – at San Francisco, CA
Here's my 5-hour conversation with @dylan522p and @natolambert on DeepSeek, China, OpenAI, NVIDIA, xAI, Google, Anthropic, Meta, Microsoft, TSMC, Stargate, megacluster buildouts, RL, reasoning, and a lot of other topics at the cutting edge of AI. This is was a mind-blowing, super-technical, and fun conversation. Yes, we discuss r1 and o3-mini, but more importantly we look into the future of technology, geopolitics, and humanity in a world that stands on the precipice of a global AI revolution. The first 4 hours are here on X (4 hours is current limit), and the full 5 hours are up everywhere else. Links in comment. Timestamps: 0:00 - Introduction 3:33 - DeepSeek-R1 and DeepSeek-V3 25:07 - Low cost of training 51:25 - DeepSeek compute cluster 58:57 - Export controls on GPUs to China 1:09:16 - AGI timeline 1:18:41 - China's manufacturing capacity 1:26:36 - Cold war with China 1:31:05 - TSMC and Taiwan 1:54:44 - Best GPUs for AI 2:09:36 - Why DeepSeek is so cheap 2:22:55 - Espionage 2:31:57 - Censorship 2:44:52 - Andrej Karpathy and magic of RL 2:55:23 - OpenAI o3-mini vs DeepSeek r1 3:14:31 - NVIDIA 3:18:58 - GPU smuggling 3:25:36 - DeepSeek training on OpenAI data 3:36:04 - AI megaclusters 4:11:26 - Who wins the race to AGI? 4:21:39 - AI agents 4:30:21 - Programming and AI 4:37:49 - Open source 4:47:01 - Stargate 4:54:30 - Future of AI
46
47
9
923
167,033
Pumped for @w4nderlus7 and team!
Today, I’m excited to unveil a project that’s incredibly close to my heart. As a lifelong gamer, I’ve always dreamed of pushing the boundaries of what’s possible in video games using my expertise in AI. Over the past year, my team at @StudioAtelico and I have been blending the magic of gaming with AI to create something special and groundbreaking: an on-device AI engine for games that empowers NPCs, agents, and natural language interactions in ways that were previously impossible or prohibitively expensive. To showcase its potential, we built GARP, a tech demo inspired by Generative Agents. GARP demonstrates how our AI engine runs in real time, locally on-device—all without relying on the cloud and for free. Can’t wait to see what game developers will build with the Atelico AI Engine! 🚀 Details and links below 👇
1
144
Lots of exciting updates from @VirtueAI_co and their partnership with @OpenAI. The team will be at @NeurIPSConf at Booth #44, check them out!
This Post is from an account that no longer exists. Learn more
3
305
Are codegen models truly secure? Dive into the latest insights below 👇
Excited to introduce SecCodePLT🛡️: a unified platform for evaluating security risks in code generation AI! Since summer, we’ve been building a comprehensive tool to assess AI models' potential for insecure coding and facilitating cyberattacks. 🧵1/👇
1
2
1,667
Ajay Rayasam retweeted
we've opened up the private beta for Sonic on-device Sonic is the fastest cloud voice gen model (cartesia.ai/sonic), and we've squeezed all of these capabilities into a model you can run locally fill out this form if you want to get early access bcpeugvkng2.typeform.com/to/…
Replying to @cartesia
📲Sonic On-Device: The Sonic model you know and love, running on device in private beta. It’s the first ultra-realistic voice model of its kind to support real-time streaming on devices.
1
4
25
4,390
This team ships fast, excited to see new use cases emerge. Check out the blog below and sign up. Happy to connect anyone to them.
Today, we’re unveiling a significant milestone in our journey toward ubiquitous artificial intelligence: AI On-Device. Our team pioneered a radically more efficient architecture for AI with state space models (SSMs). Now, we’ve optimized and deployed them at the edge. We believe the future of AI runs on your device, where it can process continuously and is reliable, private, and secure. Read our full blog post here: cartesia.ai/blog/on-device
6
566
Proud to release Codestral-Mamba 7B with @tri_dao and @_albertgu: - Strongest code model for its size, perfect for copilot apps - Mamba architecture enables linear time inference with 256K context length - Apache 2.0 weights: huggingface.co/mistralai/mam… - Blog: mistral.ai/news/codestral-ma…
8
42
8
260
60,634
Ajay Rayasam retweeted
Release day with 2 new models 🇺🇸 Sonic English​ Improved pacing, voice cloning and pronunciation, same 135ms latency 🌎 Sonic Multilingual 6 new languages (German, French, Spanish, Portuguese, Chinese, Japanese) with a new multilingual voice library And 🩺 HIPAA compliance
7
17
5
131
18,906
Ajay Rayasam retweeted
we'll be shipping another @cartesia model next week start working with us asap if you want to get early access to all the cool stuff that's coming, this team is 🚢 new models every 2-3 weeks
8
5
1
83
10,676
Ajay Rayasam retweeted
mamba-2 is here 👀 if you want to work on bleeding edge ssms with a world class research team led by @_albertgu, we're hiring @cartesia
excited to finally release Mamba-2!! 8x larger states, 50% faster training, and even more S's 🐍🐍 Mamba-2 aims to advance the theory of sequence models, developing a framework of connections between SSMs and (linear) attention that we call state space duality (SSD) w/@tri_dao
1
6
1
67
9,515
Ajay Rayasam retweeted
excited to finally release Mamba-2!! 8x larger states, 50% faster training, and even more S's 🐍🐍 Mamba-2 aims to advance the theory of sequence models, developing a framework of connections between SSMs and (linear) attention that we call state space duality (SSD) w/@tri_dao
9
182
22
973
119,656
Ajay Rayasam retweeted
Heads up comparison, @elevenlabs vs @cartesia Cartesia almost completes before Elevenlabs starts lol realtime SSMs are going to be -wild- congrats @krandiash @_albertgu
AINews: 29 May 2024 What if you KNEW that we may soon have models can that continuously process and reason over text/audio/video with a TRILLION token "context window"? Real time? On device? thanks to @cartesia, @krandiash, @_albertgu
18
24
3
281
56,053