Research Intern: Multimodal Human-Object Interaction

Honda Ri; San Jose (CA); On-site; USD 34,000 - 69,000; Full time

Job summary

Honda Research Institute USA in San Jose, CA is seeking a motivated research intern to advance multimodal machine learning for human-object interaction understanding. You will develop representation learning methods using egocentric video, hand pose, and language to model hand-object relationships and manipulation dynamics.

The internship emphasizes rigorous experimentation and collaboration to produce high-impact results suitable for publication at top conferences.

Qualifications

• Enrolled in PhD program in computer science, electrical engineering, robotics, or related field.
• Strong background in machine learning and deep learning with CV or multimodal focus.
• Proficiency in Python and PyTorch.
• Experience with empirical ML experiments, data preprocessing, and ablations.
• Excellent communication and research publication potential.

Responsibilities

• Conduct literature reviews and propose novel directions for H-O interaction representation learning.
• Design and implement ML methods for multimodal representations from egocentric video, hand pose, and language.
• Investigate spatial relations, hand-object contact, and temporal interactions.
• Develop pipelines for data processing, model training, and evaluation.
• Contribute to publications at leading conferences (CVPR, ICCV, ECCV, NeurIPS, ICLR).
• Communicate results and collaborate with researchers to publish high-impact work.

Skills

Python

PyTorch

Machine learning

Deep learning

Computer vision

Multimodal learning

Representation learning

Video understanding

Vision-language models

Hand pose

Education

Ph.D. in CS/EE/Robotics or related

Tools

PyTorch

Python

Job description

Honda Research Institute USA in San Jose, CA is seeking a motivated research intern to advance multimodal machine learning for human-object interaction understanding. You will develop representation learning methods using egocentric video, hand pose, and language to model hand-object relationships and manipulation dynamics.

The internship emphasizes rigorous experimentation and collaboration to produce high-impact results suitable for publication at top conferences.