Research Scientist: Multimodal Learning for Embodied ...

Honda Research Institute USA, Inc.; San Jose (CA); On-site; USD 90,000 - 130,000; Full time

Job summary

Honda Research Institute USA, Inc. is hiring a Research Scientist in San Jose, CA to advance machine learning for embodied intelligence. The successful candidate will develop and evaluate models for tailored solutions, contributing significantly to AI systems.

Qualifications include Ph.D. or M.S. in relevant fields and experience with modern AI tools. This role offers a collaborative environment aimed at innovative outcomes.

Qualifications

• Ph.D. or M.S. in Computer Science, Machine Learning, AI, Robotics, or related field.
• Deep understanding of modern machine learning and deep learning techniques.
• Experience with foundation models, multimodal learning, or embodied AI systems.
• 1 - 3 years of relevant work experience.

Responsibilities

• Develop and advance machine learning models including foundation models.
• Design model architectures for unified perception and reasoning.
• Train and evaluate models on large-scale datasets.
• Conduct empirical analysis of model behavior under changing conditions.

Skills

Modern machine learning techniques

Deep learning techniques

Model training and evaluation

Multimodal learning

Agent-based experimentation frameworks

Education

Ph.D. or M.S. in Computer Science, Machine Learning, or related field

Tools

AI-assisted development workflows

Job description

Job Number: P25F11

Honda Research Institute USA (HRI-US) is seeking a Research Scientist to push the frontiers of machine learning for embodied, real-world intelligence. This is a research-focused role aimed at building systems that learn from diverse data at scale, generalize robustly to novel environments, and adapt continuously as conditions evolve. The successful candidate will develop multimodal and foundation-model-based approaches that unify perception, reasoning, and decision-making for embodied AI systems.

San Jose, CA

Key Responsibilities

• Develop and advance machine learning models, including foundation models and multimodal systems that reason over vision, language, action, and other embodied signals.
• Design model architectures that unify perception, reasoning, and decision-making for embodied intelligence.
• Train, fine-tune, and evaluate models on large-scale, diverse datasets to improve robustness, generalization, and real-world performance.
• Develop rigorous evaluation methodologies, benchmarks, and metrics for complex and potentially safety-critical AI systems.
• Build reusable research infrastructure, including evaluation suites, analysis pipelines, and experimental frameworks.
• Conduct empirical analysis of model behavior, generalization, failure modes, and adaptation under changing environments or distribution shifts.
• Contribute to publications, patents, and research prototypes that demonstrate technical impact.

Minimum Qualifications

• Ph.D. or M.S. with equivalent experience in Computer Science, Machine Learning, Artificial Intelligence, Robotics, or a related field.
• Deep understanding of modern machine learning and deep learning techniques.
• Experience training and evaluating models at scale.
• Experience with foundation models, multimodal learning, or embodied AI systems.
• Proficiency with modern research tooling, including AI-assisted development workflows and agent-based experimentation frameworks.
• Strong ability to formulate research problems, execute experiments, analyze results, and communicate findings clearly.
• 1 - 3 years of relevant work experience.

Bonus Qualifications

• Expertise in one or more of the following areas:
  • Continual and adaptive learning: lifelong learning, online learning, and adaptation under distribution shift.
  • Data-centric AI: curriculum learning, data selection, active learning, and methods for measuring and improving data quality.
  • Memory and long-horizon reasoning: persistent memory architectures, external memory systems, long-context modeling, and temporal abstraction.
  • Mechanistic interpretability: understanding and analyzing model internals, behaviors, and failure modes.
• Strong publication record in leading machine learning, AI, computer vision, robotics, or NLP venues such as NeurIPS, ICML, ICLR, CVPR, ICCV, ACL, CoRL, RSS, or ICRA.
• Interest in building robust embodied intelligence systems that can operate reliably in real-world environments.

Desired Start Date: 7/6/2026

Position Keywords: Machine Learning, Embodied Intelligence, multimodal systems, Memory and long-horizon reasoning