Scaleway | Fitting Without Training With Tabular Foundation Models

What does it mean to “fit” a model without training it? Juan Pablo Fuquen, AI/ML Engineer at Scaleway, unpacks how tabular foundation models use labeled examples as context to make predictions on new data, without retraining for each dataset. Through a live end-to-end demo, he’ll walk through data preparation, context selection, held-out validation, prediction, and interpretability. Along the way, he’ll explore how choosing the right examples shapes accuracy and compute costs, why “no training” still requires engineering, and where context limits and inference costs constrain these models in practice.