For years, tabular data has been the unsung engine of AI. While transformers dominated headlines and vision models captured imaginations, tabular foundation models were steadily revolutionising how we work with structured data.
The field is exploding right now. Papers are dropping weekly, methods are evolving rapidly, and we believe this progress deserves a dedicated space where researchers can gather, share, and build together. That is why we are starting TabTalks.
We are not just organising talks. We are building a community. This is a recurring gathering where the tabular AI and time-series research community can come together to share findings and engage in the kind of deep technical conversations that push the field forward.
Every two weeks on Thursdays, we bring together researchers, students, and faculty at our offices to hear from a guest speaker. The format is simple: a 45-minute presentation followed by a 15-minute Q&A.
We are kicking things off on April 2, 2026, with Vasilii Feofanov, who will introduce Mantis, a lightweight, calibrated foundation model for user-friendly time-series classification.
Find all the details and sign up here.
These sessions are designed to nurture the research community, so expect a high technical level.
If you are working on tabular AI or time-series research and want to present your findings, please reach out. We are looking for speakers who are doing interesting work and want to share it with an engaged, technical audience.
Unlocking Predictive AI for your structured data
