Totally
Tabular

Machine learning for tabular data

Browse projects →

Projects

Have an idea? Join the Discord and tell us all about it!

Chronax ↗

Statistical and neural forecasting models under a unified, scikit-like API
Built on JAX: compiled and vectorized, so the same code runs on GPUs and TPUs for fast fitting across many series at once
Compare the precision of each model's forecasts with native conformal prediction

paper: arXiv:2604.16719

TempusBench ↗

Benchmarks foundation models against classical ones on 48 forecasting tasks
New datasets that foundation models weren't pretrained on
Every model tuned with the same harness, so comparisons are fair

paper: arXiv:2604.11529

Blog

The newest posts from the community's Substack. Want your article featured? Join the Discord and let us know!

Tabular Foundations

Oct 26, 2026 · Totally Tabular · 2 min read

Who are we?

Read on Substack ↗ Subscribe ↗

source: built-in copy · updated Oct 26, 2026

About Us

Totally Tabular is an open-source community for anyone interested in machine learning on tabular data. We offer a space for experts, students, and hobbyists alike to read, discuss, and build.

We believe that great research is the result of many diverse perspectives aligned toward a common goal. Our discussion-based reading group works through papers together, ensuring that everyone is comfortable with the literature. Community interest will determine the projects, articles, and research that we take on.

Join us on Discord, sign up for the Total Tabloid, or read us on Substack. We hope to see you soon!

Who are we?

Staff

Xan Carey

Xan is a graduate of NYU's Courant Institute of Mathematics, Computing, and Data Science. His areas of interests are machine learning interpretability and time series forecasting. At Simulacrum, he was a co-contributor to the development of Chronax, a project that eventually led him to Totally Tabular.

email: xan.carey@gmail.com linkedin: xan-carey

Staff

Anvesha Tiwary

Anvesha is a product marketer and GTM strategist who’s spent her career building marketing functions from the ground up at early-stage AI startups. It was her time at Simulacrum that got her into time-series forecasting, which is what led her to help found Totally Tabular. Here, she focuses on growing, outreach, marketing, and maintaining the community so people have a real space to dig into the research together.

email: anvesha@refrakted.com linkedin: anveshatiwary

Advisor

Deni Goktas

Deni Goktas is the founder of Simulacrum, a company that builds foundation models for time-series forecasting. Deni holds a PhD in computer science from Brown University and was a postdoctoral researcher at Cornell Tech, with research internships at Google DeepMind and J.P. Morgan AI Research. Deni's research is at the intersection of economics and computer science, and current work focuses on AI superforecasters.

email: deni@smlcrm.com linkedin: denizalp-goktas

Totally
Tabular

Projects

Chronax ↗

TempusBench ↗

Open Questions

[Open question]

Blog

Tabular Foundations

About Us

Who are we?

Xan Carey

Anvesha Tiwary

Deni Goktas

TotallyTabular

Chronax ↗

TempusBench ↗

[Open question]

Tabular Foundations

Xan Carey

Anvesha Tiwary

Deni Goktas

Totally
Tabular