I indexed 37h of my YouTube library using an RTX 4090 and local ML models in 24h

5•iliashad•11h ago

TLDR: I indexed 37 hours and 38 minutes of my videos using the self-hosted Docker version of my project on a rented RTX 4090

- Total compute time: 24 hours 13 minutes

- Frame analysis took ~85% of the time

- Heaviest plugins: TextDetection (especially on coding/screen recordings) and FaceRecognition (on podcasts)

Full article here: https://iliashaddad.com/blog/i-indexed-37h-of-my-youtube-library-using-an-rtx-4090-and-local-ml-models-in-24h

Comments

iliashad•11h ago

and the project is fully open source here: https://github.com/iliashad/edit-mind

toomuchtodo•11h ago

Does this support an S3 compatible target? If not, would you be interested in adding support for that?

iliashad•11h ago

Currently, not. Because the project is intended for local video files.

I added it as a GitHub issue: https://github.com/IliasHad/edit-mind/issues/159.

Btw, feel free to contribute to the project :)

Ask HN: Which TV do you recommend for privacy?

In LLM era, can software startup survive?

Pepperwind, a fictional company dataset with 223 tables and 79M rows

Tell HN: Scribd Breached

Heretic 2.0 Released: Training-free language model modification

I indexed 37h of my YouTube library using an RTX 4090 and local ML models in 24h

Tell HN: X is opening any tweet link in a webview whether you press it or not

Ask HN: What's it like working at Google nowadays? Core teams like Search, Ads?

Ask HN: Anyone here still writing code for reasons other than fun or learning?

Ask HN: Is applying to Y Combinator worth it if you aren't an insider?

Ask HN: Opus 5.5 suddenly dumb as rocks?

Ask HN: I'd like to request Microsoft open source Windows XP

Ask HN: Old experimental website that made photos sing?

Ask HN: Do you have side-projects that generate >$500 revenue/month?

Which hobby do you think is most diverting and pleasurable?

Ask HN: Are most of you using HN Replies? Or what else do you use?

Tell HN: Claude Code adds your email to the prompt, the LLM sends it to websites

"Wrong" Email

Ask HN: Hack

Ask HN: Is the Microsoft buyout just a long overdue correction?

Ask HN: Are you using a custom harness?

Ask HN: Are Knuth's Stanford Pages Down?

Ask HN: What's the future like? For normies, will it still be apps and websites?

GEMM Tutorial and UlmBLAS

Ask HN: Are you putting off hardware upgrades because of AI coding tools?

Ask HN: Is your leadership pushing you to skip dev review?

Ask HN: Are you using AI to do election choices like Mayor, Special Measures

Ask HN: Will apps be going away in favor of personal assistants?

Ask HN: What back end and database are you choosing for new projects in 2026?

Ask HN: Antrophic tightened usage limits?

Ask HN: Which TV do you recommend for privacy?

In LLM era, can software startup survive?

Pepperwind, a fictional company dataset with 223 tables and 79M rows

Tell HN: Scribd Breached

Heretic 2.0 Released: Training-free language model modification

I indexed 37h of my YouTube library using an RTX 4090 and local ML models in 24h

Tell HN: X is opening any tweet link in a webview whether you press it or not

Ask HN: What's it like working at Google nowadays? Core teams like Search, Ads?

Ask HN: Anyone here still writing code for reasons other than fun or learning?

Ask HN: Is applying to Y Combinator worth it if you aren't an insider?

Ask HN: Opus 5.5 suddenly dumb as rocks?

Ask HN: I'd like to request Microsoft open source Windows XP

Ask HN: Old experimental website that made photos sing?

Ask HN: Do you have side-projects that generate >$500 revenue/month?

Which hobby do you think is most diverting and pleasurable?

Ask HN: Are most of you using HN Replies? Or what else do you use?

Tell HN: Claude Code adds your email to the prompt, the LLM sends it to websites

"Wrong" Email

Ask HN: Hack

Ask HN: Is the Microsoft buyout just a long overdue correction?

Ask HN: Are you using a custom harness?

Ask HN: Are Knuth's Stanford Pages Down?

Ask HN: What's the future like? For normies, will it still be apps and websites?

GEMM Tutorial and UlmBLAS

Ask HN: Are you putting off hardware upgrades because of AI coding tools?

Ask HN: Is your leadership pushing you to skip dev review?

Ask HN: Are you using AI to do election choices like Mayor, Special Measures

Ask HN: Will apps be going away in favor of personal assistants?

Ask HN: What back end and database are you choosing for new projects in 2026?

Ask HN: Antrophic tightened usage limits?

I indexed 37h of my YouTube library using an RTX 4090 and local ML models in 24h

Comments