I am a PhD student in Aerospace Engineering at the University of Colorado Boulder, working in the Autonomous Vehicle Systems Laboratory under Dr. Hanspeter Schaub.
My research focuses on applying deep reinforcement learning to autonomous spacecraft scheduling and inspection, particularly under strict operational constraints such as power, wheel momentum saturation, data storage, and illumination geometry.
Previously, I led mission design efforts for a green orbit transfer vehicle at Infinite Orbits, conducted maneuver simulations for the Emirates Mission to the Asteroid Belt at LASP, and co-authored research on space-based inspection and safety-aware planning.
My background spans physics, aerospace systems, and machine learning — across Europe, the Middle East, and the U.S. I’m passionate about pushing the boundaries of autonomy in orbit to support next-generation space exploration missions.
Work
-
Autonomous Vehicle Systems Lab — Graduate ResearcherDeveloping Basilisk-based RL simulation for RSO surveillance under power, FOV, and pointing constraints.
Presented at AMOS 2025 and Planning & Scheduling for Space Workshop in Toulouse.
-
Laboratory for Atmospheric and Space Physics — GNC EngineerRan Monte Carlo simulations for the Emirates Mission to the Asteroid Belt (EMA).
Supported trajectory design, gravity assist performance, and interplanetary phase planning.
-
Infinite Orbits — Main Systems EngineerLed system development of GEO Ryder OTV with consortium of 8 European companies.
Designed safety concepts and FDIR architecture for a servicer docking with GEO clients for life extension.
-
ALTEN — Aerospace ConsultantManaged the Fast-Monitoring suite for ESSP, automating ground and satellite signal processing pipeline.
Oversaw GNSS receiver tracking and performance metrics across entire European service region.
-
UPC — Space Systems Project LeadLed design of wildfire-tracking satellite over Australia, with focus on instrument and AOCS design.
Developed sub-pixel resolution imaging concept to detect fires smaller than sensor resolution.
-
NYUAD — Rocketry Team Chief EngineerLed development of Level-3 solid-fuel rocket for international competition.
Oversaw propulsion, avionics, and structural integration; first UAE university to compete internationally.
-
NYUAD Galaxy Formation Group (Center for Astro, Particle and Planetary Physics) — Undergraduate ResearcherDeveloped novel mapping technique to accelerate baryonic simulations from dark matter-only data.
Reduced hydrodynamic sim times by 90%, enabling large-scale analysis of ‘missing’ dark matter structure.
-
NYUAD — Exoplanet Atmospheres Researcher
Developed ML-based data structures for modeling exoplanet atmospheres and JWST spectral profiles.
Simulated atmospheric retrieval models for different planetary mass/radius classes.
Projects
RL-Driven Scheduling for Autonomous Space-to-Space Surveillance
This project introduces a reinforcement learning (RL)-based approach for autonomous inspection and imaging of Resident Space Objects (RSOs) from a spacecraft in orbit. Unlike Earth observation missions with predictable ground tracks, RSOs exhibit dynamic, evolving motion, requiring real-time scheduling decisions by the inspecting agent.
The task is formulated as a Markov Decision Process (MDP), and the agent is trained in a high-fidelity Basilisk simulation. The RL agent dynamically selects imaging targets while managing LOS constraints, battery and reaction wheel states, and data limits. This autonomy enables persistent, priority-based surveillance without human intervention or ground-tasked commands.
As space becomes more congested, this research offers a scalable, intelligent solution to Space Situational Awareness (SSA) and Space Domain Awareness (SDA). The project extends MDP-based scheduling, previously used in Earth observation, to the domain of active space-based surveillance—significantly improving responsiveness and operational coverage.
Project Phases:
- Phase 1: Single-target RL imaging with LOS, battery, and wheel saturation constraints.
- Phase 2: Multi-RSO scheduling with dynamic, priority-based observation logic.
- Phase 3: Integration of eclipse and solar constraints for realistic observability windows.
The system is benchmarked against traditional optimization-based Earth Observation scheduling agents, evaluating imaging efficiency, resource balancing, and adaptability to dynamic conditions.
🎥 Project Video
📄 Abstract: Space-to-Space Surveillance
This project develops a reinforcement learning (RL)-based scheduler for autonomous Resident Space Object (RSO) inspection under strict safety and lighting constraints. It integrates spacecraft health management (battery, data storage, wheel desaturation) and environmental observability (eclipses, line-of-sight) into a high-fidelity Basilisk simulation.
The motivation for this work stems from challenges encountered during my time at Infinite Orbits, where I led safety concept development for the docking and life-extension of a large GEO satellite client. Precise visual navigation during the final rendezvous phase required careful handling of lighting conditions — an experience that sparked this line of research into optimizing autonomous RSO inspection with real-world constraints.
Key Features:
- Phase 1: Train RL agent for single-target imaging with LOS, battery, and wheel saturation constraints.
- Phase 2: Enable multi-target scheduling using dynamic, priority-based task selection.
- Phase 3: Integrate eclipse observability into target selection to optimize real-world inspection windows.
📄 Abstract: Autonomous Satellite Inspection
The most recent paper on this work is titled, "Autonomous Satellite Inspection in Low Earth Orbit with Optimization-Based Safety Guarantees". More research is being done to include the use of mutliple inspecting satellite working collaboratively as well as even further enhanced illumination constrained including glare and self-shadowing!
Developed a genetic algorithm capable of optimizing complex interplanetary missions using gravity assists and impulsive maneuvers.
The algorithm balances low ∆v and fast arrival time, avoiding pruning of non-intuitive solutions. Compared results with existing codes and real-world trajectories.
This research explores the development of a robust, scalable system for extracting and purifying water from subsurface resources on the Moon and Mars. Water is essential for life support, agriculture, fuel generation, and in-situ construction. A comparative analysis of technologies such as sublimation, filtration membranes, and electrochemical purification is conducted under lunar and Martian conditions.
The project culminated in a model that combines multiple methods for optimized water yield and purity, considering local regolith composition, temperature cycles, and energy constraints. The study was carried out at the Castelldefels School of Telecommunications and Aerospace Engineering (EETAC) in Barcelona, Spain.
