Research
This page contains a curated list of my major research projects, grouped by theme. Select a tag to highlight the related work. For a complete list organized by publication type, see all publications. I maintain these pages manually so they may be out-of-date; for the most up-to-date list please see my Google Scholar page.
Space Mission Autonomy
Decision making and optimization for space operations and exploration. This work spans the full operational stack of an Earth-observing or deep-space mission: scheduling image collection across large satellite constellations with maximum independent set methods and Markov decision processes, selecting cost-optimal ground station networks for downlinking that data, and building the astrodynamics tooling those algorithms depend on.
World Models for Robust Systems
By applying principles of deep learning and representation learning, we can train models that learn to predict the future state of the world based on past actions and observations. These world models can be used for planning and control in complex, high-dimensional environments with inherent quantification of uncertainty, enabling robust decision-making and control across a range of real-world robotic systems.
Decision Making for Space Safety
Application of formal decision-making frameworks, including Markov decision processes, to the operational safety of space systems and the sustainability of the space environment, with a focus on autonomous collision avoidance and space traffic management under uncertainty.
Red-Teaming Safety-Critical Systems
Automated red-teaming methods for failure discovery and validation of AI systems. By searching for the inputs that induce failure rather than waiting to encounter them in deployment, these methods surface vulnerabilities in language models and the safety-critical domains they are increasingly used in, including cybersecurity and mental health.
Other
Other research projects and publications in decision making, planning, and machine learning that do not fit neatly into the themes above.