Pierre-Yves Ritschard
@pyr
Followers
2K
Following
1K
Media
95
Statuses
5K
slacker, climber, yak shaver. founding engineer @poolsideai
Lausanne
Joined September 2009
The tech preview of LLMD is out: - Easy Setup - Just mount your model and run - Cross-Platform GPU Support - Single container works on *both* NVIDIA and AMD GPUs - Lightweight - Only 2.4GB container size - High Performance Enjoy !
15
30
175
We can finally say it publicly: @poolsideai got their models running on @awscloud’s Trainium/Inferentia thanks to @zml_ai. No code change. Promise kept.
14
26
203
Come find us next week, lots to show and to talk about :-)
We'll have a large presence at re:invent this year. Come see us!
0
0
3
.@poolsideai will be at re:Invent next week – if you’re in Vegas and want to see what we’ve been doing in the last 18 months, I’ll be on stage with Anthony Liguori at 2:30pm local time next Tuesday. More details here → https://t.co/13y6BSNumy ⛱️
0
6
30
We're 12 days out to our very first AWS re:Invent. Find us at booth #708 with demos, and exclusive merch – come say hi! https://t.co/3BOXiPI8NS
0
6
12
New @poolsideai site is incredible — awesome job @almonk, @dizzyup, and co 👏🏻 Built with Tailwind of course 💅🏻
5
11
265
oh hey we raised some money and have a new website
poolside.ai
We are software's leading AI research lab.
33
5
260
ZML: a high-performance AI inference stack that can parallelize and run deep learning systems on lots of different hardware. It's out of stealth, impressive, and open source.
26
232
2K
Attention everyone, we wrote another book 🚨 On... writing quality tech blog posts! There's also a promo code if you're really really into preordering books: `mldunlop` Enjoy! https://t.co/7ju4Cvw18O
1
10
68
FoundationDB is a major - and relatively unknown - cheat code to get to reliable large scale data storage
Do you want to know what I’ve been working on for the past few years with #foundationdb? 🤩🤩🤩
0
0
3
This is pretty much where I am at. In real world situations (for example: building an insurance claims processing system that covers 15 states), I suspect that for the foreseeable future a team of five human software engineers using AI coding tools is going to build a 10-50x
If you could fully automate software engineering (my job), I think that would be great, since I could then move on to higher-leverage things. Making software is a means to an end, not the end (software is different from art in that sense). However I have not yet seen any signs
10
8
213
For me, the biggest "wow" was how powerful a technical insight can be. Ryan realized years back that S3 allows us to build fundamentally different architectures that are super efficient in completely new ways. S3 is as revolutionary as SSDs were, but it took us longer to
This interview was great. Covers the gamut from schemaless columnar stores to S3 based architectures to advice on structured logging!
0
2
30
where can ai help us the most when it comes to programming? I don’t think it’s primarily in writing code, but refactoring it, testing it, documenting it, translating it. some early glimpses of what we’re working on @poolsideai
10
3
106
we’ve been busy
Preview of what's coming in early '24 Dive in and sign up for early cabana access at https://t.co/qBxgeOse37
1
1
67
This is the correct way to use AI for programming, IMHO.
I haven't found much use for AI in full-on writing large bodies for code for me, but I continue to be astounded by its ability to be a superb pair programmer. It knows all the APIs, we never need to Google anything, and its suggestions are often delightful. A+.
8
14
169
This is a very important thing for anyone looking at IaC right now. Even more so in environments with very frequent changes. Not everything should be in Git, especially not things better stored in databases.
We didn't use IaC to get rid of ClickOps. We used IaC to get repeatability, policy enforcement, and scale - at the expense of the user experience in ClickOps. It was a good bet, and it's remarkable how well it worked out. But it's not the only bet we could have made!
0
1
2
Hilarious seeing all the Q* guesses. Some of them come very close to what we do at poolside and call Reinforcement Learning from Code Execution Feedback.
2
1
42