Podcast · 246 episodes

Latent Space: The AI Engineer Podcast

by Latent.Space · English

The podcast by and for AI Engineers! In 2025, over 10 million readers and listeners came to Latent Space to hear about news, papers and interviews in Software 3.0. We cover Foundation Models changing every domain in Code Generation, Multimodality, AI Agents, GPU Infra and more, directly from the founders, builders, and thinkers involved in pushing the cutting edge. Striving to give you both the definitive take on the Current Thing down to the first introduction to the tech you'll be using in the next 3 months! We break news and exclusive interviews from OpenAI, Anthropic, Gemini, Meta (Soumith Chintala), Sierra (Bret Taylor), tiny (George Hotz), Databricks/MosaicML (Jon Frankle), Modular (Chris Lattner), Answer.ai (Jeremy Howard), et al. Full show notes always on https://latent.space Sponsorship and business inquiries: business@latent.space www.latent.space

Science Technology

Follow in VO

Latest episodes

Claude Code’s Next Era — Thariq Shihipar, AnthropicThariq Shihipar (Anthropic) discusses the “next era” of Claude Code/agentic coding: how the harness has evolved (CLI → agents → artifacts → CloudTag/Projects → CloudMods), why “whiplash” happens as…29 Sep 2026 · 1 h 32 min

OpenRouter: from Seed to Stripe — with OpenRouter’s Alex Atallah & AMP’s Anjney MidhaOpenRouter’s approach to model “pub-sub” and routing—why a marketplace-like layer is needed to make many LLMs usable, discoverable, and swappable for both developers and agents.25 Sep 2026 · 1 h 21 min

Runway’s WorldPrompt and the Engineering of Real-Time WorldsA conversation with Runway about how generative video tools evolved into real-time “world models,” covering Runway’s technical stack (from early image/video generation and VFX rotoscoping to Gen…25 Sep 2026 · 1 h 36 min

All episodes

🔬Bio-security is an AI Arms Race - Eric Nguyen (CEO, Radical Numerics)Eric Nguyen (Radical Numerics) explains “genetic language models” (GLMs) and their evolution from hyenaDNA (long-context DNA reading up to ~1M bases) to Evo/Evo2 (generative genomics) and Omni (a…23 Sep 2026 · 1 h 32 min · 36 chapters

🔬 An Oscar, Two Asteroids, and the Algorithm in Your sklearn: John Platt on AI for ScienceAI for Science at Google—how “Latents-Based Science” (Aira) turns scientific problems into “scorable tasks” and uses Gemini to iteratively generate and mutate code to maximize a scoring function,…22 Sep 2026 · 2 h 1 min · 53 chapters

Jev: System One models for Prod, not God — with Diogo Almeida, CEO, TypeSafe AILaunch of TypeSafe AI’s “Jev” (System One / machine-native, large-programmable models) aimed at “build prod, not god.” Jev is positioned as intelligence-per-dollar frontier software infrastructure,…21 Sep 2026 · 2 h 21 min · 63 chapters

Underwriting Superintelligence: Backing Agents you can Sue — Rune Kvist, AIUCAIUC (AI Underwriting Company) raises $40M to build “confidence infrastructure” for frontier AI—standards, third-party testing, and insurance—so enterprises and governments can adopt agents/models…16 Sep 2026 · 1 h 26 min · 37 chapters

Humanity’s Last Invention — Richard Socher of RecursiveRichard Socher (Recursive) explains the “Eureka Machine” idea: a goal-directed superintelligence that can invent most things for humanity, plus how Recursive aims to build “recursive…14 Sep 2026 · 1 h 32 min · 38 chapters

🔬“We have foundation models for language, not for physics” — Anima Anandkumar, Bren Professor of ComputingAI for science—how to build reliable “foundation” models for physical systems using neural operators (especially Fourier Neural Operators), plus formal verification with TorchLean (neural nets…26 Aug 2026 · 1 h 24 min · 30 chapters

Simulation: the new Scaling Law — Joon Sung Park, Simile AISimulation as a “new scaling law” for modeling human behavior, to enable counterfactual planning (not just prediction) and eventually large-scale multi-agent “time machine” societies.21 Aug 2026 · 1 h 10 min · 25 chapters

🔬The BioAI Phase Shift - Matthew McPartlon & Neil Patil, Chai DiscoveryChai Discovery’s BioAI “phase shift” from slow, expensive, trial-and-error protein/antibody discovery toward a more loop-like, agile workflow using structure prediction and design models, plus how…11 Aug 2026 · 1 h 35 min · 40 chapters

The Inference Engineering Masterclass — Philip Kiely & Ali Taha, BasetenInference engineering for production LLMs at Baseten—how long-context queries are routed and executed on GPU, how speculative decoding and KV-cache disaggregation improve tokens/sec, when to switch…3 Aug 2026 · 1 h 41 min · 38 chapters

Codex from 0 to 10M Users: Building ChatGPT Work — Akshay Nathan, OpenAIAkshay Nathan (OpenAI) explains how ChatGPT Work merges Codex and ChatGPT into a single “super app” for knowledge work, driven by model capability, shared “harness” primitives, and enterprise…28 Jul 2026 · 1 h 9 min · 34 chapters

Inside the Model Factory — Eiso Kant, Poolside AIPoolside AI’s approach to building open foundation models via an end-to-end “model factory” (data pipelines, distributed training reliability, and post-training/agent workflows), plus why they moved…23 Jul 2026 · 1 h 55 min · 51 chapters

🔬Causal Models Need Causal Data - Xaira’s X-Cell model for Drug Discovery (Bo Wang & Ci Chu, Chief Discovery Officer & Chief AI Scientist)Zara Therapeutics’ X-Cell “virtual cell” model for causal drug discovery, emphasizing that causal prediction needs causal perturbation data (not just observational single-cell data).21 Jul 2026 · 1 h 30 min · 37 chapters

🔬 The Lab of the Future Should Feel Like a Data Center — Andy Beam & Rafa Gómez-Bombarelli, Lila SciencesLila Sciences’ “AI science factories” for post-training at scale, arguing the lab of the future should operate like a data center—dense, energy-efficient, and instrumented for rapid, flexible…16 Jul 2026 · 1 h 41 min · 47 chapters

Why AI Infrastructure must evolve for Agent Experience — Akshat Bubna, Modal CTOModal CTO Akshat Bubna explains why AI infrastructure must evolve from developer experience to “agent experience,” arguing that Kubernetes-style orchestration is a poor fit for bursty, specialized…8 Jul 2026 · 58 min · 24 chapters

🔬 The Coolest Diffusion Research Isn't in LLMs — Evan Feinberg & Sergey Edunov, Genesis Molecular AIGenesis Molecular AI’s approach to protein–small molecule interaction modeling, emphasizing diffusion-based 3D complex prediction at near-angstrom accuracy to improve potency/ADME-relevant…1 Jul 2026 · 1 h 49 min · 49 chapters

Why the Frontier Ecosystem must be Open — Matei Zaharia and Reynold Xin, DatabricksDatabricks’ push for an open “agent cloud” ecosystem (Omnigene) and how open protocols enable interoperability; plus the “LTAP” (lakehouse HTAP) vision for unified transactional + analytics workloads.24 Jun 2026 · 1 h 9 min · 29 chapters

Red-Teaming after Mythos — Zico Kolter & Matt Fredrikson, Gray SwanGraySwan discusses “red-teaming after Mythos” (indirect prompt injection and jailbreak robustness) and how to secure AI coding/browser agents by treating models as untrusted.22 Jun 2026 · 1 h 6 min · 31 chapters

The Professor of Outputmaxxing — Anjney Midha, AMPOutputmaxxing for AI infrastructure—how to reduce GPU/data-center waste via better utilization targets, iterative bring-ups, and “compute pooling” through an independent system operator model (AMP).18 Jun 2026 · 59 min · 27 chapters

🔬 The Self-Driving Lab — Joseph Krause, Radical AIRadical AI’s “self-driving lab” approach to materials discovery for alloys, emphasizing that AI must go beyond predicting composition to also capture experimental synthesis, characterization, and…17 Jun 2026 · 1 h 17 min · 36 chapters

Reality: The Final Eval — Lukas Petersson and Axel Backlund of Andon LabsAndon Labs’ “Reality: The Final Eval” discussion of long-horizon agent evals and their vending-machine business benchmarks, plus real-world “Project Vending” deployments and multi-agent governance.4 Jun 2026 · 1 h 16 min · 41 chapters

🔬Scaling Past Informal AI - Carina Hong, Axiom MathVerified AI for “scaling brilliance” via formal verification and Lean.3 Jun 2026 · 1 h 33 min · 47 chapters