Isaac Flath
BuildPodcastCourseConsultingSubscribe

Podcast

Upcoming

Join the conversation live

The next conversation hasn’t been announced yet.

Latest episode

Earlier episodes ↓

Wednesday, November 4, 2026 · 37 min

Why is Speculative Decoding Fast?

With Isaac Flath and Joe Barrow

Why speculative decoding can make an LLM faster while doing more computation, and how GPU memory bandwidth, batch size, and draft acceptance affect the speedup.

Episode details & notes →Watch on YouTube ↗

Episode archive

Earlier conversations

Wednesday, November 18, 2026 · Recording not available

Unit Test Benchmarks

Unit tests can make strong benchmarks. Joe Barrow and I will discuss why they are useful, when they are a good fit, what their results can and cannot tell you, and how we would design them.

→

Wednesday, October 28, 2026 · 55 min

Decision Models: Inside Perplexity Decider

What decision models are, how to call them, and how Perplexity Decider works and was built.

→

Wednesday, October 21, 2026 · 31 min

Document Layout with PP-DocLayoutV3

A recorded reading and discussion of the RT-DocLayout paper with Isaac Flath and Joe Barrow, exploring PP-DocLayoutV3’s boxes, reading order, and segmentation.

→

Wednesday, October 14, 2026 · Recording not available

OCR Post-Processing and MinerU-Popo

Joining tables and paragraphs across pages, organizing headings, and deciding when the extra processing helps.

→

Follow all my work by email.

An email when I have something worth sending. Unsubscribe anytime.

BuildPodcastCourseConsultingAboutContactInstagramYouTubeThreadsXLinkedInPrivacy
© 2026 Isaac Flath