/areas

Six active research areas. Each documents the current landscape, open problems, and a running analysis of the work.

Representation Learning

active

Learning structured, transferable internal representations from data — the foundation everything else builds on.

explore →

Continual Learning

active

Learning sequentially without forgetting — making models that accumulate knowledge over time the way humans do.

explore →

World Models

active

Learning internal models of how the world works — compressing environment dynamics into something an agent can plan with.

explore →

Reinforcement Learning

active

Learning to act through interaction — the study of sequential decision-making under uncertainty.

explore →

Causality & Tabular Foundation Modeling

active

Causal reasoning meets tabular-scale foundation models — understanding and predicting structured, heterogeneous real-world data with causal guarantees.

explore →

Multimodal AI

active

Learning across modalities — vision, language, audio, time-series, and beyond. Open exploration of how AI can see, hear, read, and reason across any domain simultaneously.

explore →

These areas form a coherent cluster: representations are the substrate; world models apply them to environment prediction; RL uses them for decision-making; continual learning asks how to retain them over time. Causal inference provides the reasoning framework. Multimodal AI is where these converge at scale.