Fernando Martínez-López
I'm a PhD student in Computer Science at Fordham University, supervised by Juntao Chen.
I study reinforcement learning, representation learning, and world models. I connect representation learning and control with game theory and optimization. My current focus is decentralized multi-agent learning using world models.
My first-author work includes SCORER (ICLR 2026) and CORAL (IJCAI 2026).
I'm looking for a research internship in reinforcement learning for summer 2027.
Selected research
Stackelberg Coupling of Online Representation Learning and Reinforcement Learning
ICLR, 2026
A Stackelberg formulation lets representations adapt faster than value estimates, stabilizing their interaction in deep Q-learning.
In-Context Reinforcement Learning via Communicative World Models
IJCAI, 2026
A pretrained world model communicates reusable context, helping a separate control agent learn faster and adapt to new tasks.
Independent Multi-Agent Reinforcement Learning with Counterfactual Semantic-Social World Models
arXiv preprint, 2026 (Submitted manuscript)
Pretrained world models predict the consequences of alternative actions, improving fully decentralized on-policy multi-agent learning.
Ask the Expert: LLM-Guided Reinforcement Learning for Autonomous Cyber Defense
IEEE GLOBECOM, 2026 (Accepted; proceedings forthcoming)
LLM advice shapes rewards during training, improving sample efficiency for autonomous cyber defense without LLM calls at deployment.
Experience
May – Jul 2025
Fordham–IBM Research Intern
Fordham University
Collaborative research on representation learning and reinforcement learning using Stackelberg games and bilevel optimization. First-author work published at ICLR 2026.
Jul 2023 – Aug 2024
Quantitative Modeling, Assistant Vice President
MUFG Americas · CCAR Capital Planning
Developed quantitative models and automated an interconnected capital planning model suite, with validation, performance monitoring, and stress testing.
Sep 2021 – Jul 2023
Data Scientist
Smart Energy
Built end-to-end machine learning models and ran A/B experiments to support marketing campaigns for more than 10,000 warm leads.
Outside research
I studied jazz guitar at the Santo Domingo Conservatory for three years. I'm still a jazz nerd, and these days I'm also learning chess.
You can reach me at f.martinezlpz@gmail.com.