Matthew Braga

speaker

322 appearances 6 recordings 1 series first heard Jun 2026 last heard 11 Aug

Matthew Braga’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 6 in all, peaking in Jun 2026 with 3.

Appearances

newest first · ▶ plays the moment

Working Smarter · Building AI that can search inside videos (and photos and audio too) · 14 Jul 2026 podcast

Like you you gave that Jurassic Park example where, you know, there's scene changes, there's different people on screen, you know, later on when there's dialogue introduced, there's different people speaking, audio cues.

Like there's all of these different factors that you also have to take into account.

And I'm wondering how many of these different things are you are you looking at when when you're examining scenes?

This process of Pooh's describing, where you have a model that's been trained on a bunch of data to recognize certain patterns, and then you use that model to identify those same patterns in new data, is called inference.

Of course, I'm making this sound a lot easier than it actually is.

Who you really want to hear from is Hicham Bhadri.

He's the person responsible for helping Dropbox do inference as quickly and efficiently as possible.

Why is search and content understanding so much more difficult with things like video and audio and images compared to text?

Like why is it so hard to make that content searchable?

Can you tell me more about how many different models we're talking about here?

Like i in in Dropbox's context, for example, how many different models are are we using to try and understand some of those complex relationships you're talking about?

I'm wondering if you can tell me a little bit about how you are able to do this work so quickly when it comes to search and understanding?

Like how is it that someone can ask a question and you can find the answer within an hours long video in seconds?

And they're doing that in advance.

They're they're assuming some assumptions about what it is that people might want to know about that video or that piece of content.

I I guess that's one of the advantages of being able to go and also look at external information as being able to to then actually identify what kind of dog it is.

Like if you need to search the web for, you know, different types of dog breeds or something.

Okay.

So we don't even need to do that.

Showing 121–140 of 322 · page 7 of 17