About
I am a PhD student in the Department of Computer Science at Carnegie Mellon University (CMU). I am fortunate to be advised by Professor Tuomas Sandholm. I am also supported by the NSF Graduate Research Fellowship.
My research interests center around autonomous decision-making in high-stakes domains, and the associated technical and practical challenges of these settings. I enjoy tackling the full spectrum of open questions related to building more intelligent autonomous agents—from theory to algorithms to implementations and deployment.
Some more general areas of interest:
- Multi-Agent Systems
- Planning, Scheduling, and Operations
- Game Theory
- Combinatorial Optimization
Prior to joining CMU, I received my BSc. in computer science from McGill University. I then worked for two and a half years as a technical group member of the Artificial Intelligence Group at NASA's Jet Propulsion Laboratory (JPL). At JPL, I had the opportunity to research, develop, and deploy autonomous systems alongside many brilliant people. Most notably, my supervisor, Steve Chien.
Research
Current Projects
My current research focuses on game theory, optimization, social choice, and operations as well as leveraging AI in heart transplant allocation for the United Network for Organ Sharing (UNOS).
Selected Previous Projects
Multi-Agent Satellite Systems: Distributed autonomy can enable Earth-orbiting satellites to optimize their observations to perform time-critical and coordinated tasks such as tracking natural disasters. We work on this distributed autonomy for decentralized control of Earth-observing satellites. This work leverages distributed constraint optimization and automated scheduling to optimize spacecraft observations, resources, and time. We examine both fully coordinated settings and, more practically, federated systems. This research culminates in the FAME project, which will be the largest in-space demonstration of multi-agent AI to date.
Dynamic Targeting: Dynamic Targeting (DT) optimizes spacecraft observations and resources by using information from a lookahead sensor to identify targets for a primary sensor to seek or avoid. Deploying DT relies on synthesizing computer vision and automated planning while handling limited resources, uncertainty, and a continuous state space. The first flight demonstration of DT took place in July 2025 aboard the CogniSAT-6 Spacecraft. Other variations of DT include the VISTA project.
Publications
Dynamic Distributed Constraint Optimization and Metareasoning for Continual, Large-Scale Satellite Operations
Preprint 2026; extension of AAMAS 2026 extended abstract
|
PDF
I Zilberstein, S Chien
Aligning Data-Driven Predictors with Allocation: A Decision-Focused Approach to Survival Analysis
Preprint 2026
|
PDF
I Zilberstein, I Anagnostides, T Sandholm
Learning Potentials for Dynamic Matching and Application to Heart Transplantation
Preprint 2026
|
PDF
I Zilberstein, I Anagnostides, ZW Sollie, A Kilic, T Sandholm
Position: Machine Learning for Heart Transplant Allocation Policy Optimization Should Account for Incentives
ICML 2026, (Spotlight, top 5%)
|
PDF
I Anagnostides*, I Zilberstein*, ZW Sollie, A Kilic, T Sandholm
Near-Optimal Dynamic Matching via Coarsening with Application to Heart Transplantation
ICML 2026
|
PDF
I Zilberstein*, I Anagnostides*, ZW Sollie, A Kilic, T Sandholm
Large-Scale Continual Scheduling and Execution for Dynamic Distributed Satellite Constellation Observation Allocation
AAMAS 2026, (Extended abstract)
|
PDF
I Zilberstein, S Chien
Is Four Enough? Automated Reasoning Approaches and Dual Bounds for Condorcet Dimensions of Elections
AAMAS GAIW 2026
|
I Zilberstein, RE Berker, GZ Li, R Martins
Dynamic Targeting of Satellite Observations Using Supplemental Geostationary Satellite Data and Hierarchical Planning
ICRA 2026
|
PDF
A Kangaslahti, I Zilberstein, A Candela, S Chien
Decentralized, Decomposition-Based Observation Scheduling for a Large-Scale Satellite Constellation
JAIR 2025; earlier version in ICAPS 2024
|
PDF
I Zilberstein, A Rao, M Salis, S Chien
Real-Time Instrument Planning and Perception for Novel Measurements of Dynamic Phenomena
ASTRA 2025
|
PDF
I Zilberstein, A Candela, S Chien
Dynamic Targeting - Flight Report
ASTRA 2025
|
PDF
S Chien, I Zilberstein, A Candela, D Barretta, D Rijlaarsdam, T Hendrix, A Dunne, A Perrocheau, CC Traba, OC Grau, A Gol, MP Bove, O Aragon, JP Miquel
Federated Autonomous Operations: A New Paradigm for Large-Scale Observation Systems
SpaceOps 2025
|
PDF
I Zilberstein, A Candela, S Chien
Flight of Dynamic Targeting on CogniSAT-6 - Update
SpaceOps 2025
|
PDF
S Chien, I Zilberstein, A Candela, D Rijlaarsdam, A Perrocheau, A Dunne, T Hendrix, OC Grau, AG Mestre, MP Bove, O Aragon, JP Miquel
Multi-Asset New Observing Systems Flight Demonstration
SpaceOps 2025
|
PDF
S Chien, I Zilberstein, D Barretta, D Rijlaarsdam, T Hendrix, A Dunne, OC Grau, AG Mestre, O Aragon, JP Miquel
Passive Interception, Decoding and Processing of Public Broadcast L-Band EO Data in LEO
ESA Workshop on Advanced Flexible Telecom Payloads 2025
|
F D'Anzeo, T Dell, M Rodegari, S Chien, A Candela, Q Yue, M Kurowski, I Zilberstein
Rapid Multi-Mission Deployment of Convolutional Neural Network and Spectral Algorithm Flight Software
Flight Software Workshop 2025
|
I Zilberstein, A Candela, S Chien, D Rijlaarsdam, T Hendrix, L Buckley, A Dunne, V Vatsal, A Kothandhapani, A Subramanian, H Vyas, A Jayaswal, VS Purohit
Demonstrating Onboard Inference for Earth Science Applications with Spectral Analysis Algorithms and Deep Learning
ISAIRAS 2024
|
PDF
I Zilberstein, A Candela, S Chien, D Rijlaarsdam, T Hendrix, L Buckley, A Dunne
Learning-Based Planning for Improving Science Return of Earth Observation Satellites
ISAIRAS 2024
|
PDF
A Breitfeld, A Candela, J Delfa, A Kangaslahti, I Zilberstein, S Chien, D Wettergreen
Flight of Dynamic Targeting on the CogniSAT-6 Spacecraft
ISAIRAS 2024
|
PDF
S Chien, I Zilberstein, D Rijlaarsdam, T Hendrix, A Dunne, O Aragon, JP Miquel
Leveraging Commercial Assets, Edge Computing, and Near Real-Time Communications for an Enhanced New Observing Strategies (NOS) Flight Demonstration
IGARSS 2024
|
PDF
S Chien, A Candela, I Zilberstein, D Rijlaarsdam, T Hendrix, A Dunne
Dynamic Targeting Scenario to Study the Planetary Boundary Layer
IGARSS 2024
|
PDF
A Candela, JD Victoria, I Zilberstein, M Kurowski, Q Yue, S Chien
Search Applications for Integrated Planning and Execution of Satellite Observations using Dynamic Targeting
ICAPS PlanRob 2024
|
PDF
A Kangaslahti, I Zilberstein, A Candela, S Chien
Boolean Functions with Small Approximate Spectral Norm
Discrete Analysis 2024
|
PDF
TM Cheung, H Hatami, R Zhao, I Zilberstein (α-β)
Honors
ICON Award for AI:
for dynamic targeting SpaceNews, 2025.
NSF Graduate Research Fellowship.
National Science Foundation, 2025.
Voyager Award:
outstanding achievement in research of multi-agent AI. Jet Propulsion Laboratory, 2024.
Dean's Convocation Prize:
outstanding academic achievement throughout the B.Sc. program. McGill University, 2023.
First-Class Honors in Computer Science:
highest distinction for graduating students. McGill University, 2023.
Faculty of Science Undergraduate Research Award:
research award under the supervision of Professor Hamed Hatami. McGill University, 2022.
Dean's Honor List:
awarded based on academic GPA. McGill University, 2020-2022.
Wing Hing Chan Scholarship in Science:
awarded on the basis of academic merit. McGill University, 2021.
Tomlinson Engagement Awardee for Mentoring:
enhancing undergraduate science education. McGill University, 2020.
Faculty of Science Scholarship:
awarded based on academic achievement. McGill University, 2020.
James McGill Scholarship:
recognizing accomplishments and contributions to society. McGill University, 2019.
Valedictorian:
highest distinction for graduating students. Amherst-Pelham Regional High School, 2019.
