Alejandro Saucedo | KubeCon 2025 Europe AI Keynote
LightNews LightNews
banner
axsaucedo.bsky.social
Alejandro Saucedo | KubeCon 2025 Europe AI Keynote
@axsaucedo.bsky.social
65 followers 13 following 270 posts
Zalando Director of Eng, Science, Product & Analytics | Advisor at the UN, EU Commission, ACM, Institute for Ethical AI & others
Posts Replies Media Videos
axsaucedo.bsky.social
In 2025, about 23% of organisations are training using Databricks to train their ML models (vs 18% in 2024), followed by 13% using AWS Sagemaker (vs 14% in 2024):

The survey results for the 2025 State of Production ML: ethical.institute/state-of-ml-2025 🚀🚀🚀
December 8, 2025 at 6:03 PM
axsaucedo.bsky.social
The recent DeepMind documentary is actually a must-watch! I remember the good old days working in AI back in the mid 2010s, watching how AlphaGo blew minds with move 37...

Documentary: www.youtube.com/watch?v=d95J...

#ML #MachineLearning #ArtificialIntelligence #AI
December 8, 2025 at 11:01 AM
axsaucedo.bsky.social
Sebastian Raschka has put together the best breakdown of DeepSeek's recent V3.2 model release, and there are some pretty interesting architectural choices that make it clear why it's so far ahead:

Overview: magazine.sebastianraschka.com/p/technical-...
Paper: huggingface.co/deepseek-ai/...

#ML
December 7, 2025 at 6:03 PM
axsaucedo.bsky.social
In production ML we are reminded that your AI systems can only be as robust as the engineering discipline behind them; what better way to polish your foundations than with the Advent of Code for 2025:

Advent of Code: adventofcode.com/2025/

#ML #MachineLearning #AI
December 2, 2025 at 6:03 PM
axsaucedo.bsky.social
Sebastian Raschka and Andrej Karpathy's educational material is becoming a key cornerstone for foundational ML knowledge; and Raschka's recent course on "Building an LLM (From Scratch)" is becoming a must-watch classic:

Course: www.youtube.com/playlist?lis...

#ML #AI
December 2, 2025 at 11:02 AM
axsaucedo.bsky.social
In 2025, about 43% of organisations take between 1-3 months to productionise a machine learning model (vs 2024 28%) - and about 12% take between 3-6 months (vs 25% in 2024):

Add your response: forms.gle/KF16EckuxNUK... 🔥
December 1, 2025 at 6:03 PM
axsaucedo.bsky.social
Last week we saw a major reminder to us ML practitioners on software supply chain attacks, this one infected ~500 packages including major ones from Zapier, ENS, AsyncAPI, PostHog, Postman and others that amount to 100M+ monthly downloads:

Blog: www.aikido.dev/blog/shai-hu...

#ML #AI
December 1, 2025 at 11:00 AM
axsaucedo.bsky.social
Some of you may remember the release of the FLUX model from Black Forest Labs last year taking image generation to a whole new level; this week they have now released the 2.0 version:

Blog: bfl.ai/blog/flux-2
Technical Paper: bfl.ai/blog/flux-2
December 1, 2025 at 9:46 AM
axsaucedo.bsky.social
DuckDB is great for local development. But what if you want to take DuckDB to distrbuted data-lake scale? This is a great video on how DuckLake tackles this:

Video: www.youtube.com/watch?v=YQEU...

#ML #MachineLearning #ArtificialIntelligence #AI
November 25, 2025 at 6:04 PM
axsaucedo.bsky.social
The MCP protocol is a living standard that continuously evolves through SEPs (like PIPs in python); this week a new SEP introduces standardised UI interfaces for MCP servers which seem to be a potential game-changer in usability:

MCP SEP: blog.modelcontextprotocol.io/posts/2025-1...

#ML #AI
November 25, 2025 at 11:01 AM
axsaucedo.bsky.social
MLFlow is dominating Experiment Tracking space in ML in 2025 with 57% adoption (vs 42% in 2024)! Additionally, it makes me very happy to say that this year, Spreadsheets are down to 3% (vs 10% in 2024) - great work everyone 😂! You can add your response: forms.gle/KF16EckuxNUK...
November 24, 2025 at 6:03 PM
axsaucedo.bsky.social
We continue to see exciting waves in the field of GenAI World Models (aka text to 3D navigable environments); this week META releases a new system for AI generation of 3D worlds:

Blog: www.meta.com/en-gb/blog/w...

#ML #AI
November 24, 2025 at 11:00 AM
axsaucedo.bsky.social
These last years there's been exciting innovations in weather (+ climate) forecasting with foundation / deep learning models; Google DeepMind just released WeatherNext 2.0 which introduces a probabilistic forecaster

Blog: blog.google/technology/g...
Paper: arxiv.org/abs/2506.10772
November 24, 2025 at 8:24 AM
axsaucedo.bsky.social
Time-series forecasting quietly runs the modern world, and the race for Forecasting Foundation MOdels is growing! TempoPFN tackles univariate zero-shot forecasting with a linear RNN backbone with purely synthetic data:

TempoPFN: arxiv.org/abs/2510.25502

#ML #AI
TempoPFN: Synthetic Pre-training of Linear RNNs for Zero-shot Time Series Forecasting
Foundation models for zero-shot time series forecasting face challenges in efficient long-horizon prediction and reproducibility, with existing synthetic-only approaches underperforming on…
arxiv.org
November 21, 2025 at 11:00 AM
axsaucedo.bsky.social
What happens when you train a model using keyboard inputs to predict images of environments? You end up with interactive 3D AI worlds!

Marble: www.worldlabs.ai/blog/marble-...
NeuralOS: neural-os.com
LeCunn Update: www.nasdaq.com/articles/met...

#ML #MachineLearning #AI
November 21, 2025 at 7:53 AM
axsaucedo.bsky.social
Vector Databases are still skyrocketing: About 56% of orgs reported using vector databases, with 14% using PineconeDB (+9% YoY), about 11% have custom built in-house tools, and 8% use Azure AI (cosmos/search) surprisingly!

Add your input: forms.gle/KF16EckuxNUK... 🔥
November 19, 2025 at 11:00 AM
axsaucedo.bsky.social
The OWASP Foundation has now released the Top 10 Vulnerabilities for 2025!

Although this is focused for application security, these are also critical in production machine learning systems!

OWASP Top 10: owasp.org/Top10/2025/0...
Agentic Reports: genai.owasp.org/resources/

#AI
November 18, 2025 at 6:04 PM
axsaucedo.bsky.social
Stanford CS236 is one of the best in-depth courses on Generative Deep Learning (LLMs, TTS, VAEs, GANs, etc) and provides a super comprehensive deep dive into the key foundational concepts that are now becoming the backbone of modern AI products.

Playlist: www.youtube.com/watch?v=XZ0P...
November 17, 2025 at 11:00 AM
axsaucedo.bsky.social
Exciting news in the machine learning ecosystem this week! Today we dive into the Stanford University Course on Deep Generative Models, the OWASP Foundation Top 10 for 2025, GenAI Foundation Models from World Labs + more 🚀

lnkd.in/eRBQzVcA

#ML #MachineLearning #AI
November 16, 2025 at 6:03 PM
axsaucedo.bsky.social
Agentic systems are becoming ubiquitous, and the PyTorch team enters the race with a new agentic framework TorchForge powered by their new distributed actor framework Monarch:

TorchForge: pytorch.org/blog/introdu...
Monarch: pytorch.org/blog/introdu...

#ML #AI
November 11, 2025 at 6:03 PM
axsaucedo.bsky.social
We are excited to release another preview for our Survey on Production MLOps!! Model Serving is still unstandardised in 2025: About 40% of orgs still run custom wrappers (vs 56% in 2024) for in their ML models! Add your response: forms.gle/KF16EckuxNUK...
November 11, 2025 at 11:01 AM
axsaucedo.bsky.social
For better or for worse, AI-assisted engineering is fast becoming a core capability; DX released their 2025 report with important insights:

Report: getdx.com/uploads/ai-a...
November 10, 2025 at 6:03 PM
axsaucedo.bsky.social
We've entered the ImageNet moment for Tabular Data! A new Tabular Foundational Model is released with TabPFN 2.5 tackling zero-shot predictions; and the sky is the limit in a world that runs on tabular data (spreadsheets anyone?):

Paper: storage.googleapis.com/prior-labs-t...
November 10, 2025 at 11:01 AM
axsaucedo.bsky.social
In case you missed it last week, I share a curated set of resources from the Machine Learning Engineering ecosystem every week! In this #360th edition I dive into the new Tabular Foundation model from Prior Labs + more 🚀

www.linkedin.com/pulse/issue-...

#ML #AI
November 9, 2025 at 6:03 PM
axsaucedo.bsky.social
This is a fantastic example of "learning PyTorch the Hard Way", but often one of the best ways which is by diving into the internals after something doesn't work as expected:

Blog: elanapearl.github.io/blog/2025/th...

#ML #MachineLearning #ArtificialIntelligence #AI
November 5, 2025 at 11:01 AM