Researcher compresses culinary knowledge into 2MB model

Josef Chen built a compact machine learning model that captures cooking techniques and recipes in just 2 megabytes—small enough to run locally without cloud dependencies. For makers building cooking apps or food-tech products, this demonstrates how domain-specific knowledge can be distilled into efficient, deployable models that don't require expensive inference infrastructure.