Javier de la Rúa Martínez

🎓 Doctoral student at @dcatkth | AI Systems // 🚀 Research engineer at @hopsworks // 🥋 Kyokushinkai

[Hopsworks @hopsworks]

🚀 What if your AI platform could build pipelines, run models & host applications in the same environment?
This is where the data stack is heading → coding-agent-powered AI platforms.

📅 Live demo on May 27 👇
https://t.co/l0V12iB7aG
#MLOps #AIInfrastructure #ClaudeCode

[Hopsworks @hopsworks]

If you're spending more time maintaining ML infrastructure than building models, something’s wrong.

Hopsworks SaaS gives you the full AI lakehousefeature store, pipelines, model serving. Free tier for your first project.

Go build 👉 https://t.co/Ul3dgZoxRB

[Jim Dowling @jim_dowling]

Enough with the dunking on Europe's tech companies.
Europe's leading E-Commerce company, @Zalando, is building it real-time AI personalization on @hopsworks from Stockholm. They couldn't build it on others.

Put that in your pipe and smoke it!

https://t.co/hAp5vhsOzY

[Hopsworks @hopsworks]

🚀 1,600+ registrations, 13+ sessions — free & online.
Feature Stores + Real-Time AI with Uber, Lyft, Pinterest, Zalando & more.
Register: https://t.co/A6Zc42hbAk
Can’t join live? Just register to get all recordings.

[Hopsworks @hopsworks]

Hello everyone!
We’re excited to announce the 5th edition of the Feature Store Summit bringing 10+ of the world’s leading engineering teams to share how they build infrastructure for AI, ML, and real-time systems. Register now for free: https://t.co/A6Zc42gDKM

[Antonis Kouzoupis @alafroiskiotos]

Curious how @hopsworks Public Key Infrastructure has evolved over the years? Read in this article how this core component has been adapted to a #Kubernetes first-class citizen.
https://t.co/Kb6ANRL66G

[Pau Labarta Bajo @paulabartabajo_]

Join 3k+ members to the 𝗦𝗲𝗿𝘃𝗲𝗿𝗹𝗲𝘀𝘀 𝗠𝗟 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝘁𝘆 🤗

A Discord community of ML builders 👩‍💻👨🏽‍💻 focused on building 𝗿𝗲𝗮𝗹-𝘄𝗼𝗿𝗹𝗱 𝗠𝗟 𝗮𝗽𝗽𝘀
↓↓↓
https://t.co/qFNB5UXZv7

[Jim Dowling @jim_dowling]

This is the easiest way to install Hopsworks 4.x on Kubernetes (all clouds and on-premises k8s supported).

[Javier de la Rúa Martínez @Javierdlrm]

Looking forward to presenting at @pydatanyc later this week!

[Hopsworks @hopsworks]

Only a few days left until @PyData NY 🔥

We’re proud to be able to present two talks this year! In @Javierdlrm’s session he’ll describe the capabilities that need to be added to a Lakehouse to make it an AI Lakehouse. 

https://t.co/ZLUP20x6mI

[Hopsworks @hopsworks]

At the Feature Store Summit, @Javierdlrm presented how Hopsworks supports the snowflake schema data model. ❄️

The snowflake schema data model allows you to reduce the number of serving keys that your AI application need to make a prediction. 

https://t.co/Z6mIxcTcGn

[Hopsworks @hopsworks]

Throwback to PyData Berlin and @Javierdlrm's demo on how to build a personalized Bitcoin (BTC) virtual assistant in Python. Javier uses Hopsworks and LLM function calling to do so.

https://t.co/L2gXCJVw8Y

[Jim Dowling @jim_dowling]

The main benefit of the #Lakehouse is openness. We have Apache projects for table formats, batch/sql and streaming query engines, and catalogs.
The OSI layered architecture is a huge success. Could you imagine if the transport layer required auth/access-control by a vendor?

[Hopsworks @hopsworks]

From London to Amsterdam! 🇳🇱🌷

@Javierdlrm and @rvanbruggen represented Hopsworks at @pydataamsterdam this week, keeping participants happy by handing out Hopsworks beers and chocolates. Thank you to everyone who listened to our talk or stopped by the booth!

[Rik Van Bruggen @rvanbruggen]

My friend @Javierdlrm came up with a great analogy. Lots of people end up building their own custom #AI and #ML pipelines, but they often end up like a fragile house of cards. If you want to build great production AI/ML #systems, you need @hopsworks (and, in the picture, our "mlhops" as well) to support you.

How good is that?

[Rik Van Bruggen @rvanbruggen]

While we get ready for @Javierdlrm 's talk at @pydataamsterdam, I will just have to restrain myself and not touch either the @hopsworks chocolate or the beer. It's a tough life.

[Rik Van Bruggen @rvanbruggen]

Next week we are going to be at #pydata #amsterdam with @hopsworks . Super looking forward to lots of great discussions on our booth, and of course to @Javierdlrm's talk on thursday afternoon. Hope to see you there! (https://t.co/FdwpBjtEmn)

[Jim Dowling @jim_dowling]

Hopsworks im ComputerWeekly. Als ich in Koeln in den 1990er Jahren woehnte war ich Abonnent. ComputerWeekly ist immer noch da.

https://t.co/wKvQVSWRD7

[Daniel D. Gutierrez @ddgutierrez73]

Introducing The AI Lakehouse - https://t.co/nYVUtpx5b8

#AI #LakeHouse @hopsworks

[Pau Labarta Bajo @paulabartabajo_]

Training ML models is easy.

Transforming the data these models need is the hard part... until you learn this ↓↓↓

𝗧𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺

Building a real-world ML system is

> 𝟭𝟬% about training and deploying ML models,
and
> 𝟵𝟬% about transforming the data these models needs to work.

And the thing is, 𝗻𝗼𝘁 all data transformations are the same.

𝗧𝗵𝗲 𝘁𝗮𝘅𝗼𝗻𝗼𝗺𝘆 𝗳𝗼𝗿 𝗗𝗮𝘁𝗮 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻𝘀

In every ML system we can have up to 3 different types of data transformations

1️⃣ 𝗠𝗼𝗱𝗲𝗹-𝗜𝗻𝗱𝗲𝗽𝗲𝗻𝗱𝗲𝗻𝘁 Transformations, for example rolling averages.
> Reusable across models
> Stored in the feature store

2️⃣ 𝗠𝗼𝗱𝗲𝗹-𝗗𝗲𝗽𝗲𝗻𝗱𝗲𝗻𝘁 Transformations, for example feature normalization
> Specific to one model
> Applied in both training and inference

3️⃣ 𝗢𝗻-𝗗𝗲𝗺𝗮𝗻𝗱 Transformations
> Require real-time data
> Used in online inference

Once you understand how and 𝗪𝗛𝗘𝗥𝗘 your data transformation happens, you are in a good position to start building ML software that works.

If you want to learn more about the taxonomy of data transformations read this excellent blog post by the great @jim_dowling

> 🔗 https://t.co/q7gTBCbxag

----
Hi there! It's Pau Labarta Bajo 👋

Every day I share free, hands-on content, on production-grade ML, to help you build real-world ML products.

𝗙𝗼𝗹𝗹𝗼𝘄 𝗺𝗲 so you don't miss what's coming next

[Jim Dowling @jim_dowling]

From our SIGMOD'24 paper, an easier read on the work we have done and are doing on making #rondb the database for real-time AI applications. If you thought #redis is good enough for real-time AI, please read this and tell us if we have changed your mind. https://t.co/YCClK5lKIi