Episode

Innovating at Scale: How LinkedIn Reinvented Optimisation with PyTorch and DNS-AID

Podcast
Open Source News Elements
Published
Jun 1, 2026
Duration seconds
590
Processing state
not_requested
Canonical source
https://cdn.element.fm/468db265-34ed-4145-94a7-fc5a234303b8/2a5ee9fc-bb17-4a57-a961-426ec4d59af5/08984d9a-2ec7-4f9c-b587-812373f1f4e5/audio.mp3
Audio
https://cdn.element.fm/468db265-34ed-4145-94a7-fc5a234303b8/2a5ee9fc-bb17-4a57-a961-426ec4d59af5/08984d9a-2ec7-4f9c-b587-812373f1f4e5/audio.mp3
JSON
/v1/public/podcasts/open-source-news-elements-7412645/episodes/innovating-at-scale-how-linkedin-reinvented-optimisation-with-pytorch-and-dns-aid
Markdown
/podcast/open-source-news-elements-7412645/innovating-at-scale-how-linkedin-reinvented-optimisation-with-pytorch-and-dns-aid.md

Actions

  • POST https://stenobird.com/v1/public/podcasts/open-source-news-elements-7412645/episodes/innovating-at-scale-how-linkedin-reinvented-optimisation-with-pytorch-and-dns-aid/transcription-requests
    Idempotently request low-priority transcript generation for this episode.
  • GET https://stenobird.com/podcast/open-source-news-elements-7412645/innovating-at-scale-how-linkedin-reinvented-optimisation-with-pytorch-and-dns-aid.md
    Read the agent-friendly Markdown representation of this episode resource.

Summary

In this episode of Open Source News Elements, John delves into the impressive transformation at LinkedIn, where traditional linear programming struggled to keep pace with the platform’s massive optimisation needs. Discover how LinkedIn rearchitected their distributed solver, DuaLip, on PyTorch, unlocking rapid, GPU-powered speedups, while reducing engineering overhead and powering real-time personalisation for millions. Plus, learn about DNS-AID, the innovative protocol poised to radically reshape how AI agents discover and communicate across the web. Tune in for a deep dive on scaling, modernising optimisation, and advancing AI interoperability.