- Invented a machine-learning-based constellation reference governor that tunes each target orbit to trade instantaneous and lifetime-maximum formation error, reducing simulated max. error by >85% from conventional methods for no extra fuel cost- Developed multiple new local constellation formation error analysis techniques to address practical requirements that were not addressed by simpler global analyses, e.g. optical communication terminal range and ground coverage- Architected an automated constellation-scale Ground-Software/GNC interface managing spacecraft tracking, collision avoidance, formation maintenance, and orbit transfer guidance, eliminating the cost of 3rd-party solutions- Led research, development, and operation of novel Rendezvous & Proximity Operations (RPO) algorithms based on nonlinear optimization and model predictive control, resulting in autonomous rendezvous and passively safe formation flying on orbit- Developed a novel optimization-based guidance strategy enabling spacecraft rendezvous under thrust constraints (electric propulsion) and energy constraints (small battery) that could not be met by existing heritage algorithms- Created a data-driven intersatellite distance prediction method for advance (12-200 hrs) RPO planning, reducing error of close approach start time predictions by 64%-75% from pre-existing first-principles methods- Won >$1B in U.S. Gov. (Space Development Agency) and commercial contracts (on teams of 20-70) through diplomatic & responsive customer engagement, clear & thorough technical communication, and attention to detail in technical mission design contributions- Contributed to Δv budgeting, orbit perturbation analysis, and design of propulsion, launch, orbit transfer, collision avoidance, & stationkeeping from low-Earth through geostationary orbits, resulting in 4 successful proposal and design review closures

- Studied control, modeling, and sensing of repetitive processes via machine learning, hybrid systems theory, computer vision, and hands-on development of electrohydrodynamic jet printing (e-jet), a microscale resolution additive manufacturing technology mostly used for printed microelectronics.- Designed & executed experiments comparing solvents for conductive nanoparticles and substrate surface treatments to balance conductive ink cohesion and ink/substrate adhesion, and to prevent e-jet nozzle clogging (i.e. increase e-jet uptime).- Designed & executed experiments comparing toolpaths for substrate nanopositioning stages and signal shaping of the jet actuation voltage to maximize print resolution while maintaining required print conductivity.- Integrated experiment results and automated multi-instrument coordination to achieve rapid fabrication of custom microelectronics (4 point probes and interconnects for nanowire transistors), resulting in a successful DARPA seedling kickoff.- Mentored 1 PhD student and 1 master's student in microscale additive manufacturing operations, dynamic systems modeling & analysis, computer vision, and process automation, resulting in streamlined centralized control of custom microelectronics fabrication and a published paper in the IFAC/AACC 2021 Modeling, Estimation, and Control Conference.

- Studied control, modeling, and sensing of repetitive processes via machine learning, hybrid systems theory, computer vision, and hands-on development of electrohydrodynamic jet printing (e-jet), a microscale resolution additive manufacturing technology mostly used for printed microelectronics.- Created the first ODE-based model of the complete e-jet process by hybridizing physics-driven and data-driven partial models, enabling study of model-based analysis & control.- Developed a closed-form mathematical representation of piecewise defined systems, enabling the synthesis and application of traditional optimization and iterative learning control (ILC) algorithms to these nontraditional systems.- Co-invented a framework for Model Predictive Control (MPC) of Priced Timed Automata (a type of discrete event system) by integrating MPC principles with cost-optimal reachability analysis; validated in simulation on a manufacturing system.- Extended a classical computer vision technique for identifying the corners of an object by combining contextual image information with an iterative filtering process to achieve the location of dull corners in noisy, low-resolution images.- Integrated Levenberg-Marquardt-Fletcher adaptive optimization algorithms with ILC, improving optimization robustness to system uncertainty and nondifferentiability.- Initiated research on nonlinear sparse symbolic regression techniques for increasing the applicability of data-driven models.- Won multiple awards founded in communication skills, including the $1500 Towner Prize for Outstanding Engineering Graduate Student Instructors and the Best Presentation Award at the 2019 American Control Conference Special Rapid-Fire Session.

- Invented a new iterative learning control (ILC) scheme enabling the first-ever validations of model-based ILC using nonlinear models with unstable inverses, including overhead crane and printhead positioning system models.- Derived conventional inversion and stable inversion formulas for piecewise affine (PWA) system models (e.g. systems with varying contact conditions), enabling feedforward control and ILC of PWA systems with both stable and unstable inverses.- Mitigated the problem of PWA inverse model nonuniqueness by inventing an inverse model approximation technique, vastly expanding the portfolio of PWA models usable for feedforward and iterative learning control.

Hybrid Powertrain Controls Group.- Streamlined memory-intensive scripts to accelerate fuel economy analysis for large data sets, with automatic report generation.- Designed experiments and analyzed measured data to determine the root causes of customer complaints and to quantify the effectiveness of countermeasures.- Authored two bilingual technical reports, combining graphical communication and concise English and Japanese to convey complex results to speakers of either language.- Analyzed the separability and importance of many possible predictors for engine shock, resulting in scripts to automate and visualize preliminary analyses for large data sets.

- Directed a team of 4 undergraduates in the development of dynamic sensing and analysis devices for aquatic physical fitness.- Designed a sensor package enabling wireless communication to a submerged device, reducing assembly steps from 4 to 1, and adding an extra measurand versus prior models.- Upgraded an Arduino/MATLAB filtering and estimation program for real time implementation by resolving fatal communication latency and data loss problems.

- Studied computation-based design of photonic crystals for application as thermophotovoltaic emitters.- Designed a 1-dimensional photonic crystal filter for thermophotovoltaic emitters that improved emissivity suppression of undesired wavelengths by 49% and emissivity unidirectionality by ~30% in simulation versus classical designs (quarter-wave stacks).- Initiated new research based on a personal discovery using the band edges of double period 1-dimensional photonic crystal filters to selectively amplify substrate emissivity.

- Negotiate with campus officials for off-site IT services and on-campus laboratory space.- Organized team presence at college department events, increasing recruitment and connecting with potential sponsors.- Presented team mission and organization at college department meetings to increase faculty-student relations on the team.- Secured new faculty mentors through personal networking and presentation at department meeting.

- Won $4000 from the "Big Ideas @ Berkeley" competition (2nd place of over 120) by leading a team of 10 undergraduates in writing a grant for the development of a transverse flux motor for lightweight vehicles.- Implemented a new motor controller calibration method resulting in increased max torque output.- Testing and analysis of electrical vehicle motors, coordination of technical experts, negotiation with suppliers, communication with business and other technical teams.- Design and execute experiments to identify vehicle system parameters and improve models.

- Designed curriculum to teach CAD and fabrication skills to ~100 college freshmen in one month, resulting in faster time-to-contribution for new students and reducing recruit drop-out compared to previous years.- Fabricated a unique chassis design via machining, creating unique jig geometries, and mig welding.