Profile avatar
martin.elstner.dev
AI and ML engineering. Search and recommendation. Founder of Elstner Analytics. Helping companies solve their data issues https://elstner.dev
24 posts 1,202 followers 781 following
Regular Contributor
Conversation Starter

In-place Assistants > Chat windows! Hugging Face's integration of an "AI Query" overlay in their SQL console exemplifies this. Users input natural language, AI suggests SQL queries—streamlining data exploration seamlessly. Probably the best showcase of this pattern in a freely accessible product.

It’s a really nice view out of our kitchen window. But I underestimated the effect on daylight amount in winter before moving to the north. #norway

This looks promising! So far, I run my own evaluation code, because the pre-made solutions were a bit obscure for my taste. This could change my mind.

What better first post than to advertise this position. Come join our group! Our group is recruiting a postdoctoral researcher for a project on reactive machine learning potentials. Reposts and spreading the word to interested people in your network appreciated! jobs.ethz.ch/job/view/JOP...

Early in my data science career, I misread causal inference as casual inference. Like “yeah, I do inference but very casual“. Imagine my disappointment, as the text book arrived. But trying out prompts in chat interfaces brings me really close to my initial dream.

Post a picture you took (no description) to bring some zen to the timeline.

Happy to announce that Pleias is joining The AI Alliance to Co-lead the Open Trusted Data Initiative. We will release on November 11th the largest multilingual fully open dataset for LLM training with 2 trillion tokens on HuggingFace and The Alliance OTDI.

Heading to the German Conference on Cheminformatics #GCC2024. Send my poster for printing. Will showcase VLM usage for chemical data.

Please like or reply and I’ll add you to this list Trying to build an AI/ML list Also share! Trying to get everyone on here bsky.app/profile/did:...

Bluesky now has over 10 million users, and I was #764,676!

Let's try an experiment: an RDKit community survey. Basically the same survey that I used for the 2024 UGM, now I'm opening it up to the entire RDKit community. If you have a minute or two, it would be great to learn more about the community. It is, of course, anonymous. forms.gle/Z9EmfXdKQ7Y1...

Love the nuance in Kelsey‘s thread. But I have a strong bias to single node setups, especially for data engineering workloads. Many companies run smallish cluster jobs (like 8 nodes with 32 cores and 128GB ram) on <1TB datasets. This isn‘t wrong and was SOTA a few years back. But if I need to touch

I'm beyond excited to finally announce my new book "Feature Engineering A-Z" 🎉 The vision for the book is to be a comprehensive collection of feature engineering methods. Describing how they work, when and why you should and shouldn't use it. Code snippets in both R and Python! feaz-book.com