Research Intern: LLM-Enabled Multi-Agent Social Simulation

Honda Research Institute USA; San Jose (CA); On-site; USD 34,000 - 62,000; Full time

Job summary

Honda Research Institute USA is seeking a Research Intern to develop next-generation social and multi-agent simulation methods using state-of-the-art AI techniques, including large language model agents.

The internship will focus on building computational frameworks that simulate human social behavior and group interaction at scale, enabling the study of coordination, cooperation, communication, social reasoning, and norm formation.

Qualifications

• Current Ph.D. student in Computer Science, Computational Social Science, Cognitive Science, HCI, or a related field.
• Strong programming skills in Python.
• Experience with large language models, NLP, MARL, machine learning, or agent-based modeling.
• Solid background in experimental design, model evaluation, and data analysis.
• Familiarity with multi-agent systems, social simulation, computational modeling of human behavior, or human-AI interaction.

Responsibilities

• Develop LLM-agent and multi-agent simulation environments for modeling social interaction and collective behavior.
• Design agent architectures, social reasoning mechanisms, communication strategies, and interaction protocols for multi-agent social settings.
• Incorporate behavioral theory and social science constructs into simulation design, such as cooperation, communication, social influence, role dynamics, or norm emergence.
• Build evaluation pipelines to assess realism, robustness, and the plausibility of mechanisms underlying simulated behavior against human data or established theory.
• Analyze emergent behaviors and interaction dynamics in simulated populations under different social, environmental, or policy conditions.
• Conduct experiments and ablation tests on agent behavior, interaction dynamics, and simulation validity.
• Document findings in the form of technical reports, research prototypes, and publication-ready materials.

Skills

Python programming

NLP

MARL

Agent-based modeling

Experimental design

Data analysis

Education

PhD student in CS/related field

Job description

Research Intern: LLM-Enabled Multi-Agent Social Simulation - Honda Research Institute USA

Research Intern: LLM-Enabled Multi-Agent Social Simulation

Job Number: P26INT-21

Honda Research Institute USA (HRI-US) is seeking a research intern to develop next-generation social and multi-agent simulation methods using state-of-the-art AI techniques, including large language model (LLM) agents. The goal of this project is to build computational frameworks that simulate human social behavior and group interaction at scale, enabling the study of interaction dynamics such as coordination, cooperation, communication, social reasoning, norm formation, and collective decision-making. The intern will design and evaluate LLM-based agent societies and multi-agent simulations grounded in social science theory and empirical data. This work will combine LLMs, agent-based modeling, machine learning, structured models of social reasoning, and human-centered evaluation to investigate how AI agents reproduce human social behavior and group interaction, when they can serve as useful proxies for human behavior, and where they fail to capture important social dynamics. The expected outcome is a research contribution suitable for submission to top-tier venues in AI, HCI, or computational social science, with an impact on both human-AI systems and social simulation research.

San Jose, CA

Key Responsibilities

• Develop LLM-agent and multi-agent simulation environments for modeling social interaction and collective behavior.
• Design agent architectures, social reasoning mechanisms, communication strategies, and interaction protocols for multi-agent social settings.
• Incorporate behavioral theory and social science constructs into simulation design, such as cooperation, communication, social influence, role dynamics, or norm emergence.
• Build evaluation pipelines to assess realism, robustness, and the plausibility of mechanisms underlying simulated behavior against human data or established theory.
• Analyze emergent behaviors and interaction dynamics in simulated populations under different social, environmental, or policy conditions.
• Conduct experiments and ablation tests on agent behavior, interaction dynamics, and simulation validity.
• Document findings in the form of technical reports, research prototypes, and publication-ready materials.

Minimum Qualifications

• Current Ph.D. student in Computer Science, Computational Social Science, Cognitive Science, HCI, or a related field.
• Strong programming skills in Python.
• Experience with large language models, NLP, MARL, machine learning, or agent-based modeling.
• Solid background in experimental design, model evaluation, and data analysis.
• Familiarity with one or more of the following: multi-agent systems, social simulation, computational modeling of human behavior, or human-AI interaction.

Bonus Qualifications

• Prior research experience with LLM agents, generative social simulation, or agent-based modeling.
• Background in computational social science, behavioral modeling, or social science theory.
• Experience evaluating AI systems against human behavioral data, qualitative studies, or controlled experiments.
• Familiarity with prompting, tool use, memory, planning, or coordination in LLM-based agents.
• Publication record in relevant venues such as NeurIPS, ICLR, AAAI, ICML, CHI, CSCW, ICWSM, or ACL/EMNLP.
• Interest in building simulation platforms that support hypothesis generation, intervention testing, or human-AI system design.

Years of Work Experience Required 0

Desired Start Date May 2027

Internship Duration 3 Months

Position Keywords LLM Agents, Social Simulation, Agent-Based Modeling, Human Behavior Modeling, Computational Social Science