I am a Ph.D. student in the Department of Industrial Engineering and Operations Research (IEOR) at UC Berkeley, advised by Thibaut Mastrolia. Before beginning the Ph.D., I earned an M.Eng. in IEOR from UC Berkeley and a Diplôme d'Ingénieur from CentraleSupélec.
My research interests include financial mathematics, stochastic optimal control, and reinforcement learning.
I also work on AI safety with the Center for AI Risk Management & Alignment (CARMA), where I build stochastic models of cognitive firewalls for governable multi-agent AI systems.
E-mail: julius.graf@berkeley.edu
Publications
- Learning Optimal Liquidation with Closing Auctions with Thibaut Mastrolia (preprint, submitted for publication, 2026, arXiv:2601.17247)
Talks and Presentations
-
2026 INFORMS Conference on Financial Engineering and Fintech, Stevens Institute of Technology, Hoboken, NJ, USA (November 2026)
-
Workshop on "Stochastic Control and Reinforcement Learning", CIRM, Marseille, France (July 2026)
-
Minisymposium on "Advances in Market Microstructure, Market Making, and Competition", 13th World Congress of the Bachelier Finance Society, Bologna, Italy (June 2026)
Teaching Experience
I am a Graduate Student Instructor (GSI) for the following courses at UC Berkeley:
-
INDENG 222, Financial Engineering Systems I (Spring 2027)
-
INDENG 241, Risk Modeling, Simulation, and Data Analysis (Fall 2026)
I was a Teaching Assistant (TA) for the following courses at CentraleSupélec:
-
Probability, Bachelor of Global Engineering, CentraleSupélec-McGill University (2024)
-
Analysis 1-3, Bachelor of Global Engineering, CentraleSupélec-McGill University (2023-2024)
-
PDEs - Partial Differential Equations (2023-2024)
-
CIP - Convergence, Integration and Probabilities (2023-2024)