Recent Posts (all)

April 9, 2025 at 7:45 PM

What a stunning article about why uv is 10-100x faster that the competition.

Koen Leijsten takes you through all the tricks—with a style that is calm and captivating at the same time—that the fine folks at Astral used to make uv stand out as the #python package manger everyone loves!

👏👏👏

https://xebia.com/blog/uv-the-engineering-secrets-behind-pythons-speed-king/

Post image 2

Posted on 09 Apr 2025 · 5766 views · original

April 6, 2025 at 7:00 AM

Meta just dropped Llama 4—on a weekend, can we ever catch a break? 😀—and there are two open weights models, Scout and Maverick.

Scout is particularly impressive with its 10M token context (!!!), first in the industry. Quite an engineering accomplishment from Meta.

The models are already available on Fireworks AI and we started to put them through their paces—getting some impressive, Gemini-2.5-territory, performance.

Next week, we’ll set them up in our own lab—we recently added a new machine with 150GB of NVIDIA cards so we can’t wait to see how it performs on them.

The models are a big deal for self-hosted, you’re-in-control models that are customers keep on asking.

So expect a write up to understand the cost, security, and fine-tuning implications these models have for your business!

Anand Sahay Rob Dielemans Mayank Verma Preetpal Singh Klaudia Wachnio (Zdunczyk)

https://ai.meta.com/blog/llama-4-multimodal-intelligence/

Post image 2

Posted on 06 Apr 2025 · 4834 views · original

March 31, 2025 at 8:05 PM

I am honored to be a panelist at the Tech Summit Europe on Thursday, representing Xebia during the Unleashing AI: How to Compete in the AI Revolution panel.

I will be speaking about how #data and #AI separates successful companies from those that are falling behind, the biggest mistakes companies make when trying to adopt AI (hint: it’s about a lack of vision and building solutions in search of problems), and how #GenAI is being used to create a competitive advantage right now.

Get your tickets at https://techsummit.tech/europe/

Post image 2

Posted on 31 Mar 2025 · 1377 views · original

March 26, 2025 at 8:45 AM

Security debt is the worst debt you can accumulate! And, when it comes to Data & AI, it means your Data is at risk, and so is your business.

Isaac Sacolick recently interviewed me and wrote about it—and six other types of debt—on CIO Online.

And, I’m proud of Xebia team’s relentless focus on creating guardrails for LLMs that ensure only the right data reaches the model, protecting our clients and their customers!

Read the full article https://www.cio.com/article/3850777/7-types-of-tech-debt-that-could-cripple-your-business.html

Post image 2

Posted on 26 Mar 2025 · 866 views · original

March 20, 2025 at 6:59 AM

Really cool features of the new Mac Studios by Apple: with just 20,000$ of hardware, you can run the FULL DeepSeek R1 model!

https://x.com/alexocheema/status/1899735281781411907?s=42

There’s no other company in the world that allows you to run R1 so cheaply!

Post image 2

Posted on 20 Mar 2025 · 1574 views · original

March 16, 2025 at 7:29 PM

With all this LLM news going around, we forgot the most exciting thing Apple 🍎 did this year.

They renewed Ted Lasso for Season 4 ⚽️ 🥅

https://www.hollywoodreporter.com/tv/tv-news/ted-lasso-renewed-season-4-apple-tv-1236163347/

Post image 2

Posted on 16 Mar 2025 · 3456 views · original

March 14, 2025 at 4:02 PM

Called it

“Perhaps surprisingly, European banks have emerged as a hotbed of experimentation. Strict confidentiality rules limit how much the financial sector can rely on cloud-based ai services. That makes open-source models hosted internally an attractive alternative. Natwest and HSBC, two British lenders, are both experimenting with building their services on top of R1, as is Spain’s BBVA, according to the Information, a news site.”

https://www.economist.com/business/2025/03/13/western-companies-are-experimenting-with-deepseek

Post image 2

Posted on 14 Mar 2025 · 1640 views · original

March 14, 2025 at 8:47 AM

Big news in LLM land landed yesterday!

Allenai.org released a fully open-source LLM (data, code, weights, and details) model, OLMo 2 32B.

They claim it outperforms GPT 3.5 and GPT 4o mini while being relatively small. If you care about open models (why should you care? I’ve written an article about that https://articles.xebia.com/deepseeks-v3-r1-ai-smarter-cheaper-secure), you ought to try it out!

Blog post with the announcement: https://allenai.org/blog/olmo2-32B

Post image 2

Posted on 14 Mar 2025 · 2981 views · original

March 12, 2025 at 2:16 PM

Analytics at the speed of your thoughts!

MotherDuck and DuckDB just released a local UI for duckdb.

Work with notebooks, databases, and visualize the data all within the same interface.

I love it already—try it out with duckdb -ui

More info in the comments!

Post image 2

Posted on 12 Mar 2025 · 12940 views · original

March 3, 2025 at 7:55 PM

If you wanna know why I’ll never be an influencer on Instagram, look no further.

Recording the video was the most awkward thing I did all week—can’t imagine doing so for a living.

But! The webinar tomorrow will be great. You still have time to catch us live!

Posted on 03 Mar 2025 · 1904 views · original

9/72