Michael de la Maza, PhD

Announcing nanochat democratizing access to conversational AI Today, Andrej Karpathy has released nanochat, a bold new open-source project that reimagines how we build and interact with language models. At its heart, nanochat seeks to be a “strong baseline stack” a cohesive, minimal, readable, and hackable codebase that implements the full pipeline (tokenization, pretraining, fine-tuning, inference, and a Web UI) in 8,000 lines of code. Here are a few reasons why this is exciting (and why you should care): What makes nanochat stand out 1. Accessibility You can spin up a model with a single script on a cloud GPU instance and, in about 4 hours (~US$100), interact with a ChatGPT-style interface. 2. Simplicity & clarity The entire system is designed to be hackable. Anyone can fork, experiment, or extend it. 3. Scalable ambition Run it longer, and the model begins to outperform GPT-2 in benchmarks even handling math, code, and reasoning tasks. Educational potential Karpathy frames nanochat as the capstone of his LLM101n course under Eureka Labs an open lab for learning and experimenting with LLMs. Why it matters In an era where large LLMs are becoming increasingly complex and resource-intensive, nanochat is a refreshing counterpoint: a project that says, “Let’s build from first principles, with transparency, and invite everyone to learn along.” It lowers the barrier to entry for researchers, students, and curious developers to build, dissect, and deploy conversational agents reigniting the open-source AI movement. Check out the repo here: https://lnkd.in/d98x5k3k Excited to see what the community builds next with this foundation. For more such latest tech news and AI updates, follow Anubhav Dube. #AI #MachineLearning #LLM #OpenSource #AndrejKarpathy #nanochat #ConversationalAI #TechNews

4

From our PearX S25 cohort, meet Pravāh, the AI-native operating system for the electric grid When Mohak Mangal, Dhruv Suri, and Aman Gupta looked at the strain on the U.S. electric grid, they saw a system pushed to its limits. Exploding demand and unpredictable supply have made blackouts increasingly common. This costs the economy more than $150B a year, a number that could 10x in the next five years. Mohak Mangal and team founded Pravāh to help fix it. Their platform helps utilities and grid operators make real-time decisions on load, generation, and congestion, reducing blackout risks, optimizing power procurement, and bringing much-needed stability to the backbone of modern life. We’re proud to support the Pravāh team as they take on one of the toughest and most important infrastructure challenges of our time.

804

23 Comments

We believe that great companies combine outstanding deep tech capabilities and exceptional execution. That's why Alpha Intelligence Capital is partnering with Carnegie Mellon University: our team, led by Nhi Lê, will help make CMU's deep tech founders "venture-ready", in a jointly designed program spanning 6 months and starting right now. Applications are open!!! Nhi Lê Uday Sandhu Terry Chou Alberto Pelliccione Arun Vijay Arnaud Barthelemy Axel Badalian Katharina Maier Coby Yeung Alex P. Leung Arthur Paquet Charles-Edouard Bouée

22

1 Comment

Axiom, an AI startup founded by Stanford University math PhD student Carina H., is raising $50 million at a valuation of $300 million. The company aims to develop AI that can solve complex math problems for quantitative and hedge funds, despite not having launched a product yet. Carina Hong, who has a strong academic background with degrees from MIT and Oxford, said, "Math research is really fun," highlighting her passion for the field. Read more: https://lnkd.in/eY5nEV5v 📰 Subscribe to the weekly AI in Finance Newsletter: https://lnkd.in/eYWyQWiY #ai #artificialintelligence #ainews #aistartup #aifunding

5

Artificial intelligence is already reshaping biomanufacturing—challenging long-standing practices and unlocking new levels of productivity and precision. Krish W. Ramadurai from AIX Ventures, Ouwei Wang from Pow.Bio, Elliot Hershberg from Century of Biology, and Tim Dobbs from BioMADE bring deep expertise across biomanufacturing scale-up, synthetic biology infrastructure, investment strategy, and AI-native bio startups. This conversation gets past the buzzwords and into what’s really happening on the ground. The panelists will explore how AI is accelerating production timelines, lowering cost barriers, and powering smarter, more adaptive fermentation processes. At the same time, they’ll confront the hard questions—data quality, tech stack integration, and regulatory readiness—through candid insights and case studies from the lab to the plant floor. If you want an honest look at the challenges, opportunities, and future direction of applied AI in the bioeconomy, this is the #SynBioBeta session to join. Get your conference passes now: https://lnkd.in/gaEseRQa See you May 5–8 in San Jose!

32

1 Comment

🌟Supervisor Spotlight!🌟 Introducing the wonderful Bonan Zhao from School of Informatics: "My work in computational cognitive science explores how people learn, form hypotheses, and communicate. To me, responsible NLP means grounding NLP systems in how human cognition works. When people interact with language models, they rapidly form beliefs about what these systems can and cannot do—often inaccurately. My research on human learning and hypothesis formation can help us understand how users develop these mental models, and design systems that actively shape accurate expectations, preventing over-reliance, misuse, and harm. People also learn language in remarkably efficient and environmentally sustainable ways, building rich understanding from limited examples. By understanding and leveraging human inductive biases, we can create NLP systems that develop more like people do—requiring less data and fewer computational resources while achieving more meaningful generalization." Bonan Zhao's website: https://lnkd.in/e8Py9X73 Register for Applicant Webinars here: https://lnkd.in/e8B9zFC9 Find more information about the CDT ➡️ https://lnkd.in/duGN2b4Y #ResponsibleNLP #ResponsibleAI #ComputationalCognitiveScience #PhDResearch School of Informatics, The University of Edinburgh | The University of Edinburgh | Edinburgh Futures Institute | Edinburgh College of Art | 🌟

31

Axiom, a pre-product AI startup founded by a Stanford math PhD Carina Hong, is raising at a $300m valuation. Goal is to build AI trained on math proofs that can solve complex math problems with certainty, for Wall St. quants. Carina graduated from MIT with a double major in mathematics and physics. She then earned a neuroscience master’s degree with dissertation distinctions from Oxford, now pursuing a law degree and PhD in mathematics at Stanford.

64

3 Comments