Aditya "JP" Jayaprakash

speaker

248 appearances 1 recordings 1 series first heard Jun 2026 last heard 29 Jun

Aditya "JP" Jayaprakash’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 — 1 in all, peaking in Jun 2026 with 1.

Appearances

newest first · ▶ plays the moment

Uplink: AI, Data Center, and Cloud Innovation Podcast · CI Can't Keep Up With AI · 29 Jun 2026 podcast

And we quickly realized that on the hyperscalers, hyperscalers were great if you wanted to slowly add machines.

And I mean, you have EKS and you have Carpenter to help with that, but

That's not really great when you want to go from zero to, say, 8,000 cores in a matter of 10 seconds.

Wow.

So we quickly realized that both of these made it extremely hard to make it work on the hyperscalers.

8,000 cores, which is like a lot.

And in fact, we have some customers that burst all the way up to 40,000 in a matter of seconds, which is...

Absolutely crazy.

You would not be able to do that on the hyperscalers.

And we're able to handle that without breaking a sweat.

I see.

And that was really the thesis behind why we decided to build our own stack.

So that's what's sort of... Every single part of the stack, every single trade-off, every single decision that we've made is simply geared towards performance.

Absolutely.

I think pre-AI, right now we're 25 people and we're an eight person engineering team.

Pre-AI, I mean, without AI coaching tools, we'd need 100 people minimum.

So-

Something that I've observed and also talking to our customers

The amount of code that's being shipped is getting to a point where it's very hard for people to even review pull requests.

Like back in the day, there was a lot of friction in creating a PR.

Showing 121–140 of 248 · page 7 of 13