Skip to content
Deepak Gupta markPodcasts

Podcasts

Latent Space

Editorial pick

Hosted by swyx (Shawn Wang) & Alessio Fanelli

The technical AI engineering podcast, practitioners talking to practitioners.

Interview Weekly 30–60 min· ~90 episodes· Started 2023
Apple PodcastsSpotifyYouTubeRSS Show website

Editorial take

The closest thing AI engineering has to a trade publication in audio form. swyx and Alessio are practicing engineers, and it shows, questions are about real implementation tradeoffs (RAG architectures, eval frameworks, agent loops) rather than the AI-discourse meta-conversation that dominates Lex Fridman or Hard Fork. If you ship LLM-powered features for a living, this is the highest signal-to-noise show on the air.

Last hand-checked 2026-05-09.

Listen if you …

  • you ship LLM-powered features and need implementation-grade detail
  • you want to hear founders of dev-AI tools (Cursor, Cognition, Replit) on record
  • you're comfortable with terminology like 'eval set' and 'function calling' as table stakes

Skip if you …

  • you want a high-level survey of AI, this assumes you know the basics
  • you don't write or ship code, the engineering depth will be excessive

If you only listen to one

The 4 Wars of the AI Stack

swyx's mental model of where money flows in the AI stack. Useful framing every AI builder ends up referencing.

About the show

swyx and Alessio Fanelli's show on the technical and business mechanics of building with AI. Both hosts are practicing engineers, which sets the bar for guests: expect founders, infra leads, and researchers, not corporate communications people.

Episodes alternate between guest interviews and 'state of X' synthesis episodes. The newsletter that accompanies the podcast is itself worth subscribing to.

Notable guests

  • Andrej Karpathy
  • Logan Kilpatrick
  • Aravind Srinivas
  • Garry Tan
  • Jeremy Howard

Pairs with

If Latent Space works for you, these likely will too.