Ferdinando (Nando) Fioretto is an Associate Professor of Computer Science at the University of Virginia, where he leads the Responsible AI for Science and Engineering (RAISE) lab. His research develops foundational machine learning and generative AI methods for systems that must obey control objectives, hard constraints, safety requirements, and physical laws. The group works at the intersection of machine learning, generative AI, optimization, control, and responsible AI, with the goal of building models that are not only predictive or generative, but also reliable, controllable, and usable in scientific and engineering settings where feasibility matters.

A central theme of his work is the integration of learning and optimization. His group studies differentiable optimization, decision-focused learning, and learning-to-optimize methods that allow machine learning models to reason through downstream decisions and constraints during training and inference. This line of research connects classical ideas from constrained optimization, Lagrangian duality, bilevel optimization, and control with modern deep learning systems and generative AI. It has led to methods for learning fast approximations of constrained optimization problems, training predictive models for decision quality rather than only prediction error, and embedding optimization structure into generative models so that their outputs remain feasible and useful for real-world operations.

A major focus of Nando’s group is AI for science and engineering. This includes constrained diffusion and flow-matching methods, discrete diffusion models for structured generation, physics-informed learning, and neuro-symbolic approaches that combine learned models with explicit rules, simulators, and optimization layers with applications spanning protein design, molecular generation, materials discovery, robotics, power systems, scheduling, policy optimization, and other scientific and engineering domains where model outputs must satisfy geometric, physical, operational, or safety specifications.

Nando’s research also focuses on responsible AI for decision-making and learning systems, with emphasis on privacy and safety. His work has investigated how privacy-preserving mechanisms interact with fairness, how computational and systems choices can introduce disparate impacts, and how constraints can be used to improve the reliability and accountability of machine learning pipelines.