Scholar
Mahmoud Mahfouz
Google Scholar ID: FAusNh8AAAAJ
AI @ J.P. Morgan / PhD @ Imperial College London
Citations & Impact
Citations
624
H-index
6
i10-index
5
Publications
17
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- Publications: 'Entropy-Aware Branching for Improved Mathematical Reasoning' (preprint on arXiv); 'The State of the Art of Large Language Models on Chartered Financial Analyst Exams' (presented at EMNLP 2024 Industry Track); 'Capacity planning and scheduling for jobs with uncertainty in resource usage and duration' (published in The Journal of Supercomputing). Patent applications: Method and system for solving reconciliation tasks by integrating clustering and optimization; Method and system for providing dynamic workspace scheduler. Filed patent: System and method for institutional risk identification using automated news profiling and recommendation. Other achievements: Presented work on asset allocation using behavioral cloning and reinforcement learning at the AAAI 2023 Bridge Program on AI for Financial Services; Published paper 'FinRDDL: Can AI planning be used for quantitative finance problems?' in the proceedings of the FinPlan-2023 Workshop; Made available 'Towards Robust Representation of Limit Orders Books for Deep Learning Models' on arXiv, exploring the stability of LOB representations.
Research Experience
- Work experience: AI Research Lead at J.P. Morgan's AI Research lab, led by Prof. Manuela Veloso; Front-office software engineer at J.P. Morgan, developing and supporting global trader-facing software systems; Data scientist at J.P. Morgan, designing machine learning models for time-series anomaly detection and anti-money laundering screening; Internships: Software engineer at Goldman Sachs; Robotics engineer at Culham Centre for Fusion Energy (CCFE); Applications engineer at Keysight Technologies.
Education
- Degree: Master of Engineering (MEng) in Mechatronic Engineering; University: The University of Manchester; Graduated with First Class (Hons) and ranked 3rd in the Electrical & Electronic Engineering department; Advisor: Not explicitly mentioned; Field: Mechatronic Engineering. PhD: Imperial College London; Advisor: Prof. Danilo Mandic; Status: Part-time Ph.D. student.
Background
- Research interests: Development of autonomous agents for complex, multi-agent financial market environments. Focus on sequential decision-making in dynamic, non-stationary, and partially observable settings. Techniques used include deep reinforcement learning, behavioral cloning, agent-based modeling, and LLM-powered multi-agent systems. Main application areas: General decision-making problems in finance and algorithmic trading within limit order book markets, including universe selection, asset allocation, and optimal trade scheduling and execution.
Miscellany
- Personal interests: Learned a lot about remote-handling robotic systems used in the Joint European Torus (JET) nuclear fusion tokamak during an internship at the Culham Centre for Fusion Energy (CCFE).
Co-authors