CV

PhD Student

Center for Human-Compatible Artificial Intelligence (CHAI)

Advised by Dr. Stuart Russell.

• Developed theory and empirical support for a novel form of topological representation coined ‘feature fields’.
• Found emergence of world models in neural networks with statistical structure, e.g. HMMs, Ising model.
• Advised undergraduate interns on two world model projects, one leading to a workshop paper.

Master's Student Researcher

MIT Algorithmic Alignment Group

Advised by Dr. Dylan Hadfield-Menell.

• Submitted a workshop paper extending previous results from contracting in multi-agent reinforcement learning into the multi-agent generative agent domain.
• Reviewer for Foundation Models for Decision Making 2023 NeurIPS Workshop.
• Developed a new pipeline extending previous workshop results to improve negotiation with contract rollouts.

Research Intern

Center for Human-Compatible Artificial Intelligence (CHAI)

Mentored by Dr. Justin Svegliato under Dr. Stuart Russell.

• Created a novel multi-agent contracting framework for generative agents in Minecraft.
• Constructed a benchmark of social dilemmas agents to facilitate the study of multi-agent coordination.
• Demonstrated that generative agents can generate and respond to contracts in a way that overcomes social dilemmas, leading to better outcomes for all agents.

Research Intern

IonQ — Quantum Applications Team

Mentored by Dr. Daiwei Zhu.

• Designed novel patent-pending optimization techniques to make quantum circuits more efficient by training reinforcement learning agents based on graph neural network architectures.
• Developed a circuit transformation library to serve as the environment for RL agents.

Research Intern

MIT Research Laboratory of Electronics

Under Dr. William Oliver.

• Studied resonance properties of superconducting microwave resonators.
• Developed analysis software to do qubit initialization for Bosonic error correcting codes.

Research Intern

MIT Laboratory for Nuclear Science

Under Dr. Lindley Winslow.

• Published a paper on a novel multi-objective optimization technique for reconstructing particle trajectories for the Cryogenic Underground Observatory for Rare Events (CUORE).
• Designed a Monte Carlo for the CUORE underground detector to simulate particle events.
• Conducted statistical data analysis in Python and parallelized algorithms for a 50-node compute cluster.

Teaching Assistant

MIT 6.4110 — Representation, Reasoning, and Inference in AI

Worked with Prof. Leslie Kaelbling.

• Conducted weekly office hours to provide students with personalized assistance on course material.
• Lead problem sessions to reinforce key concepts and demonstrate problem-solving techniques.
• Collaborated closely with Prof. Leslie Kaelbling to ensure course materials were effectively communicated and to gather feedback for continuous course improvement.

Full Stack Developer

Tracflo (Pozen Fellowship)

Entrepreneurship internship through the MIT Pozen Fellowship.

• Designed and developed a dashboard in React and NodeJS for a startup as the centerpiece product.
• Interviewed clients to collect feedback and plan new features.

Developer

AI Bill Legislation Tracker

Built a legislation tracker for policy-makers to follow bills related to artificial intelligence in the US.

• Built infrastructure for a cloud-based database of bills with automatic updating.
• Implemented automated ChatGPT categorization and analysis of bills.
• Collaborated closely with policy researchers at CHAI to ensure the tool is as helpful as possible.

Teacher

Global Teaching Labs

Designed and taught a curriculum on quantum computing to ~30 German high school students.

• Collaborated with a student at Jacobs University to ensure the curriculum was accessible and engaging.

Communications Director

MIT AI Alignment

• Mentored three undergraduate members with career, research, academic, and personal advice.
• Organized public talks featuring prominent figures in the AI research field.
• Designed student surveys and strategized about meetings with administration.
• Facilitated weekly reading groups over two semesters, fostering collaborative discussions on the latest research.

PhD

UC Berkeley

Computer Science

• LLM Agents
• Natural Language Processing
• Physics-inspired Machine Learning
• Information Theory

Master of Engineering

Massachusetts Institute of Technology

Computer Science and Electrical Engineering

B.S. (Double Major)

Massachusetts Institute of Technology

Physics; Artificial Intelligence and Decision Making

• Machine Learning
• Deep Learning
• Reinforcement Learning
• Large Language Models
• Computer Vision
• Algorithms
• Abstract Algebra
• Topology
• Quantum Physics III
• Quantum Computation
• Relativity

Neural Manifold Geometry Encodes Feature Fields

NeurIPS 2025 Workshop on Symmetry and Geometry in Neural Representations

J. Yocum, C. Allen, B. Olshausen, and S. Russell.

Transformers Represent Causal Abstractions

NeurIPS 2025 Workshop on Symmetry and Geometry in Neural Representations

E. Altuzar, J. Yocum, and C. Allen.

Mitigating Generative Agent Social Dilemmas

NeurIPS 2023 Workshop on Foundation Models for Decision Making

J. Yocum, P. J. K. Christoffersen, M. Damani, J. Svegliato, D. Hadfield-Menell, and S. Russell.

Muon Track Reconstruction in a Segmented Bolometric Array Using Multi-Objective Optimization

Journal of Instrumentation (JINST) 17 P07004

J. Yocum, D. Mayer, J. L. Ouellet, and L. Winslow.