I am a Ph.D. candidate at NYU Tandon. My dissertation studies non-equilibrium mechanics of multi-agent learning. My work primarily centers on the intersection of statistical/reinforcement learning and algorithmic game theory. Recent applications include large language model (LLM) multi-agent security (jailbreaking analysis, agentic orchestration, fine-tuning, and red-teaming), data-driven market making and deep hedging under adversarial risk, and decentralized multi-radar interference mitigation.

Education

  • Ph.D., NYU in Electrical Engineering, 2021–2026

    • Committee: Prof. Quanyan Zhu, Prof. Elza Erkip & Prof. Sundeep Rangan
    • Thesis: Variational Foundations of Multi-Agent Learning: A Unified Non-equilibrium Theory

  • M.Sc., NYU in Electrical Engineering, 2018–2020

  • BEng., BUPT in Communication Engineering, 2014–2018

Working Experience

  • Research Intern — NXP Semiconductors, 2025

    • Developed a high-fidelity multi-agent FMCW radar simulation framework with advanced signal filtering techniques.
    • Designed decentralized no-regret algorithms for FMCW radar interference mitigation.

  • Research Intern — Augur Lab, 2024

    • Developed a signature-method-based market simulator for derivative instruments.
    • Applied general minimax-concave (GMC) regularized spline fitting for financial time-series modeling.

  • Teaching & Graduate Assistant — NYU Tandon, ECE Larx Lab, 2019–2025

    • Taught and assisted graduate-level courses in system optimization, probability, robotics, and game theory.
    • Led research projects on large language model (LLM) multi-agent security, including jailbreaking analysis, agentic orchestration, fine-tuning, and red-teaming dataset construction.

Selected Projects

  • Game-Theoretic Market Making under Adversarial Risk (2025)

    • Developed a multi-layer game-theoretic extension of Avellaneda–Stoikov market making for adversarial environments, modeling strategic drift manipulation and inventory-dependent exploitation.
    • Built a high-fidelity simulation platform incorporating stochastic volatility and regime-switching dynamics, calibrated on cryptocurrency market data.

  • Deep Hedging with Reinforcement Learning under Partial Observability (2025)

    • Designed an RL-based hedging framework for multi-asset options in partially observable markets, using approximate information states to handle latent volatility.
    • Simulated high-dimensional price paths via multi-factor stochastic volatility models.
    • Implemented RNN/Transformer deep hedging agents to achieve PnL-efficient hedging.

  • Decentralized Learning for FMCW Radar Interference Mitigation (2024–2025)

    • Developed a multi-agent radar simulation framework modeling mutual interference in dense FMCW sensing environments.
    • Designed decentralized no-regret learning algorithms enabling autonomous radars to adapt transmission strategies without coordination.
    • Demonstrated convergence to stable interference-aware operating points and significant performance gains in detection and estimation accuracy under realistic signal and noise models.

Publications

Journal Articles

  1. Y. Pan, T. Li, Q. Zhu. “Model-agnostic hessian-free meta-policy optimization via zeroth-order estimation: A linear quadratic regulator perspective,” Dynamic Games and Applications, 2025. DOI.

  2. Y. Pan, Q. Zhu. “A games-in-games paradigm for strategic hybrid jump-diffusions: Hamilton-jacobi-isaacs hierarchy and spectral structure,” 2025. Under submission to IFAC Nonlinear Analysis: Hybrid Systems.

  3. Y. Pan, J. Li, L. Xu, S. Sun, Q. Zhu. “Decentralized no-regret frequency-time scheduling for FMCW radar interference avoidance,” 2025. Under submission to IEEE Transactions on Aerospace and Electronic Systems.

  4. T. Li, H. Li, Y. Pan, T. Xu, Z. Zheng, Q. Zhu. “Meta stackelberg game: Robust federated learning against adaptive and mixed poisoning attacks,” 2024. Under submission to IEEE Transactions on Information Forensics and Security.

Conference Proceedings and Preprints

  1. Y. Pan, J. Li, L. Xu, S. Sun, Q. Zhu. “A game-theoretic approach for high-resolution automotive FMCW radar interference avoidance,” 2025. arXiv:2503.

  2. Y. Pan, Q. Zhu. “Timing-aware two-player stochastic games with self-triggered control,” 2025. arXiv:2512.18109.

  3. Y. Pan, Q. Zhu. “Bayesian holonic systems: Equilibrium, uniqueness, and computation,” 2025. arXiv:2512.18112.

  4. Y. Pan, Q. Zhu. “Extending no-regret hopping in FMCW radar interference avoidance,” ACM SIGMETRICS Performance Evaluation Review, vol. 53, no. 2, pp. 131–133, 2025. DOI.

  5. Y.-T. Yang, Y. Pan, Q. Zhu. “Preference-centric route recommendation: Equilibrium, learning, and provable efficiency,” 2025. arXiv:2504.01192.

  6. Y. Pan, T. Li, Q. Zhu. “On the variational interpretation of mirror play in monotone games,” IEEE CDC 2024, pp. 6799–6804. DOI.

  7. Y. Pan, T. Li, H. Li, T. Xu, Q. Zhu, Z. Zheng. “A first-order meta Stackelberg method for robust federated learning,” New Frontiers in Adversarial ML Workshop, 2023. OpenReview.

  8. Y. Pan, T. Li, Q. Zhu. “Is stochastic mirror descent vulnerable to adversarial delay attacks? A traffic assignment resilience study,” IEEE CDC 2023, pp. 8328–8333. DOI.

  9. Y. Pan, T. Li, Q. Zhu. “On the resilience of traffic networks under non-equilibrium learning,” IEEE ACC 2023, pp. 3484–3489.

  10. Y. Pan, Q. Zhu. “On poisoned Wardrop equilibrium in congestion games,” GameSec 2022, pp. 191–211.

  11. Y. Pan, Q. Zhu. “Efficient episodic learning of nonstationary and unknown zero-sum games using expert game ensembles,” IEEE CDC 2021, pp. 1669–1676.

  12. Y. Pan, G. Peng, J. Chen, Q. Zhu. “MASAGE: Model-agnostic sequential and adaptive game estimation,” GameSec 2020, pp. 365–384.

Book Chapters & Technical Reports

  1. T. Li, Y.-T. Yang, Y. Pan, Q. Zhu. From Texts to Shields: Convergence of Large Language Models and Cybersecurity, 2025. arXiv:2505.00841.

  2. T. Li, Y. Pan, Q. Zhu. “Decision-dominant strategic defense against lateral movement for 5G zero-trust multi-domain networks,” Network Security Empowered by Artificial Intelligence, pp. 25–76, Springer, 2024.

  3. H. Li, T. Xu, T. Li, Y. Pan, Q. Zhu, Z. Zheng. A First-Order Meta Stackelberg Method for Robust Federated Learning (Technical Report), 2023. arXiv:2306.13273.

Technical Skills

  • Mathematical Foundations: Statistical Learning • Convex Optimization • Algorithmic Game Theory • Probability & Stochastic Calculus • Information Theory • Signal Processing • Optimal Control • Reinforcement Learning

  • Tools & Platforms: NumPy • PyTorch • Gym • Pettingzoo • SUMO • R • MATLAB • PostgreSQL • LoRA • vLLM • VectorBT • QuantLib • Gcloud • AWS Bedrock

  • Supporting Capabilities: Academic Research & Presentations • Linux • Git • LaTeX

Awards and Achievements

  • 2023 — Dante Youla Award for Graduate Research Excellence, NYU Tandon

  • 2022 — Best Paper Award, GameSec Conference 2022

  • 2020 — Merit Award, Outstanding academic performance in ECE Department

Academic Service

Leadership Roles

  • Technical Session Chair — IEEE International Radar Conference 2025

  • Session Co-Organizer — Digital twin-based Driver risk-aware intelligent mobility analytics for urban transportation management, 2025

  • Peer Reviewer (2024–2025) — Automatica, Annals of Reviews in Control, IEEE CDC, ACC, INFOCOM, CNS

Invited Conference Sessions

  • INFORMS Annual Meeting (2025) — Computational Foundations of Multi-agent learning, Atlanta, GA

  • IEEE CDC (2025) — Bounded Rationality in Human-AI Decision-Making, Rio de Janeiro, Brazil

  • IEEE International Radar (2025) — Automotive Radar Principles and Challenges, Atlanta, GA

  • IEEE CDC (2024) — Networks, Games and Learning II, Milan, Italy

  • SIAM Conference on Mathematics of Data Science (MDS24) (2024) — Atlanta, GA

  • IEEE ACC (2023) — Resiliency and Privacy Throughout Networked Cyber-Physical Systems, San Diego, CA