Ramin Hasani is the co-founder and CEO of Liquid AI, and he has done what almost no one in AI has: built a real alternative to the transformer, the architecture behind ChatGPT and nearly every model you touch. Liquid AI raised a $250M Series A led by AMD, and by his own account now sits around $4.5 billion. The core idea did not come from a bigger data center. It came from the 302-neuron brain of a 2mm worm. We also get into the part most founders never put on a slide. While the whole field was stripping complexity out of neural networks to make them scale, Ramin spent seven years adding it back in, chasing an equation about how two neurons talk that had sat unsolved since 1907. He cracked it on a chalkboard at MIT, one now headed to the MIT history museum.
Ramin was born in Iran and left for Politecnico di Milano, then a PhD at TU Wien in Vienna, where modeling the brain of a worm became Liquid Neural Networks. From there: MIT, CSAIL, a 2022 Nature paper, and in March 2023 spinning the research into Liquid AI alongside Daniela Rus, Mathias Lechner, and Alexander Amini.
Our News Letter : https://substack.com/@seondnewsletter
Ramin Hassani Linkedin : https://www.linkedin.com/in/alikashani
Ardalan Javadi : https://www.linkedin.com/in/ardalanjam1369/
Farzam Hejazi: https://www.linkedin.com/in/farzam-hejazi-5b608081/
00:00:00 Introduction
00:03::24 Why Noise Is Actually a Resource for Computation
00:06:01 Where the Whole Worm Idea Came From
00:6:56 Why Ramin Chose the C. elegans Worm as the Basis of His Research
00:10:19 How Ramin Found His First Co Founder
00:18:08 The Equation Ramin and His Team Solved That Had Been Unsolved Since 1907
00:19:06 The First Mathematical Attempt in 1953 to Show Neuron Relationships 00:19:41 What Is the Problem With Differential Equation Based Machine Learning? 00:21:29 The Aha Moment Behind Liquid AI
00:22:16 The Birth of Liquid AI
00:22:46 How Sergey Levine Inspired the Name
00:24:09 How Liquid AI's Architecture Differs From Transformers
00:25:09 The Relationship Between Bias and Algorithm Scalability
00:28:05 Why Liquid Neural Networks Have an Efficiency Advantage at Scale
00:29:41 Do Scaling Laws Apply to Liquid Models?
00:31:26 Attention Is Not the Only Thing You Can Scale
00:34:08 How the Liquid Search System Works
00:34:53 Why Liquid AI Has the Lowest Memory Usage and Latency and Why It Is Hybrid
00:35:00 The Best Foundation Model for CPUs
00: 37:30 How to Close Enterprise Customers in the AI Era
00:38:29 What Vanguard Taught Ramin About Financial Strategy
00:41:58 125 Enterprises Are Now Using the Liquid AI Model
00:42:56 How to Design a Revenue Model for Enterprise in the AI Era00:43:19 How Shopify Uses Liquid AI
00:44:00 How Mercedes Uses Liquid AI With a 600 MB Model
00:45:51 The Quality of $1 of Liquid AI Revenue Compared to $1 of Anthropic
00:47:13 Why Liquid Is Going After Arm's Business Model
00:52:53 The Story Behind Another Iranian Founded AI DNA Startup That Raised $200M in Seed
00:55:47 The Story Behind Liquid AI's $290M Fundraise
00:59:30 How to Close a Seed Round
00:1:00:17 The Mastermind Behind the Fundraising
00:1:02:57 The Story Behind AMD Leading the Round
1:04:56 Why Revenue Matters So Much for Fundraising
1:05:22 Liquid AI Is Now at a $4.5 Billion Valuation
1:05:51 Ramin's Insight on Building a World Class Network
1:07:32 Ramin's Takeaways From Meeting Jamie Dimon
1:08:31 Jamie Dimon's Pitch for Liquid AI
1:08:51 How to Build an Effective Short Narrative for Your Pitch
1:10:31 Leading by Example
1:11:07 What Is the Best Definition of Intelligence?
1:14:32 Silicon Valley Is a Gossip Town
1:16:22 Why an Exponential View Matters and How Sam Altman Does It
References: Check our News Letter : https://substack.com/@seondnewsletter?r=2v0lej&utm_campaign=profile&utm_medium=profile-page
About Us : Linkedin : https://www.linkedin.com/company/secondpodcast/
Twitter : /thesecondpod
Instagram : https://www.instagram.com/thesecondpod/ #Secondpod #Second_Pod #پادکست ـ دوم #secondpod #پادکست
D'autres épisodes de "Secondpod"



Ne ratez aucun épisode de “Secondpod” et abonnez-vous gratuitement à ce podcast dans l'application GetPodcast.








