Sebastian

Professor of data management for ML at @bifoldberlin. Ex-@UvA_Amsterdam, @NYUDataScience, @Twitter intern; member of @TheASF & @EFF. Views are my own.

[Sebastian @sscdotopen]

Job Opening:

Our lab is hiring a student employee (40h/month) for the development of a new benchmark for ML engineering agents with realistic ML pipelines.

https://t.co/MS70Gzcgh5

[François Chollet @fchollet]

That's nothing, I know software engineers in big tech who were capable of this feat even before the advent of GenAI

[Sean Kulinski @seankski]

bet you’ve heard of train on test, but have you heard of test on train?

[Sebastian @sscdotopen]

Activity of the day: tricking "AI Agents" into doom-loops ;)

[Aditya Parameswaran @adityagp]

Blog Post: Looking back on the first decade as faculty (2014-2024).

I list my favorite papers from the decade, why I enjoyed working on them, and provide backstory and reflection.

https://t.co/TdzQenNt8V

[Andrew Gordon Wilson @andrewgwils]

I miss the days of being a PhD student, or postdoc. I would give almost anything to have multiple full days at a time, just to concentrate deeply and single-mindedly on open-ended research.

[Omar Khattab @lateinteraction]

I should write or record a longer piece on this at some point. But hopefully the slides will useful to someone.

Link: https://t.co/J8KRshqOIq

[Sebastian @sscdotopen]

Next time my students ask me how real-world data looks like, I will point them to this article :)

https://t.co/8CL7ESog2t

[Aditya Parameswaran @adityagp]

New research agenda we're kickstarting at Berkeley: redesigning data systems to serve the dominant workload of the future: agents!

Agentic speculation is massive, heterogeneous, steerable, and redundant: properties data systems can better support and take advantage of.

Take a look: https://t.co/rxyJG43gXb

[PVLDB @pvldb]

Vol:18 No:12 → mlidea: Interactively Improving ML Data Preparation Code via "Shadow Pipelines" https://t.co/UlgO8QCeBz

[Sebastian @sscdotopen]

@sh_reya @HamelHusain If you can, hire @sh_reya! I would if I could ;)

[Sebastian @sscdotopen]

If you are at #icml25, don't miss @o_ovcharenko's spotlight poster today at 11 a.m. PDT — 1:30 p.m. PDT at West Exhibition Hall B2-B3 #W-311. ICML link: https://t.co/ZjqeP4H5XJ

[Olga Ovcharenko @o_ovcharenko]

Thanks to all co-authors Florian Barkmann, Philip Toma, @ImantDaunhawer, @vogt_je, @sscdotopen and
@val_boeva

📄 Full paper: https://t.co/zo21ZjzuDv
💻 Code: https://t.co/nirY7Lo2vL

[Valentina Boeva @val_boeva]

Join our lab's presentations at ICML'2025 @icmlconf in beautiful Vancouver!

1. Thursday, Olga Ovcharenko (@o_ovcharenko) will present our work with @sscdotopen and @vogt_je on "scSSL-Bench: Benchmarking Self-Supervised Learning for Single-Cell Data", selected for a spotlight poster. https://t.co/qDlzl7D5rC. Paper: https://t.co/CdSfwWLCLm

2. Saturday, Marc Glettig (@GlettigMarc) will present our work on "H&Enium, Applying Foundation Models to Computational Pathology and Spatial Transcriptomics to Learn an Aligned Latent Space", selected for a poster presentation at the Workshop on Multi-modal Foundation Models and Large Language Models for Life Sciences. Paper: https://t.co/sQzgCVhCZs ICML link: https://t.co/aRbn0KqYi9

3. Saturday, I will give an invited talk about our CancerFoundation model by @Theus__A and Florian Barkmann at the Workshop on Multi-modal Foundation Models and Large Language Models for Life Sciences. Preprint to be updated soon with new results: https://t.co/AwcIryGK9p

[Sebastian @sscdotopen]

On Saturday, @o_ovcharenko will present a poster on "Towards Cross-Modal Error Detection with Tables and Images" at the the Data World workshop, which details our initial ideas on finding errors in tables by inspecting corresponding image data:

https://t.co/9VjVWWpCpW

(3/3)

[Sebastian @sscdotopen]

The DEEM Lab is at ICML this week for the first time, with two contributions!

(1/3)

[Sebastian @sscdotopen]

On Thursday, Olga will present her research on "scSSL-Bench: Benchmarking Self-Supervised Learning for Single-Cell Data". This paper is joint work with ETH Zuerich and was selected as a spotlight poster:

https://t.co/ZjqeP4H5XJ

(2/3)

[Olga Ovcharenko @o_ovcharenko]

Our paper "Towards Cross-Modal Error Detection with Tables and Images" was accepted for the DataWorld workshop at ICML'25! 🥳

Thanks to @sscdotopen!

[Sara Magliacane hiring PhDs at UvA @saramagliacane]

New PhD position at @AmlabUva on learning concepts with theoretical guarantees using #causality and #RL with me, Frans Oliehoek (TU Delft) and @herkevanhoof 💥

Deadline: 15 June

https://t.co/q4dxcRCko5

[Sebastian @sscdotopen]

@ahmadi94f Congrats, Fatemeh!

[Sebastian @sscdotopen]

We have a PhD opening in Berlin on "Responsible Data Engineering", with a focus on data preparation pipelines designed along responsibility objectives.

This is a fully-funded position at @bifoldberlin, co-supervised by @stoyanoj from NYU.

Details: https://t.co/J4xItgjtaE