Jure Leskovec
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Relational Foundation Models for Enterprise Data with Jure Leskovec
- Published
- May 21, 2026
- Duration
- 1h 6m
- Summary source
- description
- Last updated
- Jun 10, 2026
Discusses deep-learning.
Summary
In this episode, Jure Leskovec, co-founder and chief scientist at Kumo and professor of computer science at Stanford, joins us to explore two fronts of his work: AI for science and relational deep learning. We begin with AI Virtual Cell, a multiscale effort to learn data-driven representations from proteins to cells to patients using single-cell RNA-seq d…
Intelligent report
Sign in to read teasers, or upgrade to Research Pro to commission a new dossier for this episode. Learn more →
Show notes
In this episode, Jure Leskovec, co-founder and chief scientist at Kumo and professor of computer science at Stanford, joins us to explore two fronts of his work: AI for science and relational deep learning. We begin with AI Virtual Cell, a multiscale effort to learn data-driven representations from proteins to cells to patients using single-cell RNA-seq data, protein language models like ESM, and structure models like AlphaFold—without hand-encoding biology. Jure then dives into relational deep
Themes
- deep-learning