Food is far too valuable to waste, yet food waste costs the hospitality industry over $100 billion annually. In fast-paced, human-centric, less-digitalized commercial kitchens particularly, up to 20% of purchased food is wasted, often an amount equivalent to net profits. These retail-scale food production and service environments are also notoriously data-scarce, further complicating efforts to improve efficiency and sustainability. Such challenges demand innovative solutions from the fields of operations management and operations research. Having this objective in mind, my dissertation is to explore AI solutions for running a more sustainable and efficient kitchen. Based on an industrial collaboration with Winnow, we first measured the commercial and environmental impacts of using AI (Computer Vision) to record detailed transaction-level food waste in commercial kitchens and explored the potential mechanisms that drive the impacts. We then developed data-driven algorithms that leverage these new data streams to identify what is the optimal decision-making for kitchen managers, and proposed two types of inventory co-pilots to assist with human decision-making. Finally, we experimentally evaluate how human decision makers would react to the deployment of different forms of AI assistance and uncover factors influencing their effectiveness.