AI Research Intern: Wireless Foundation Models

About the role

Bosch Research in Pittsburgh is looking for an enthusiastic research intern to advance foundation models for wireless sensing and perception. The intern explores how large-scale self-supervised and multimodal learning can build generalizable representations of human activities and environments from radio-frequency (RF) signals and wireless channel measurements. The work focuses on models that learn reusable representations and transfer across tasks, environments, devices and users, with potential applications in human presence detection, activity recognition, motion understanding and fine-grained behavioral and contextual understanding.

Key responsibilities

• Develop novel foundation-model architectures and representation-learning methods for RF signals and wireless channel measurements.
• Investigate self-supervised, contrastive, generative and multimodal learning approaches for large-scale wireless sensing data.
• Study model generalization and adaptation across environments, users, devices and downstream perception tasks.
• Develop and evaluate methods for fine-tuning, few-shot learning or zero-shot transfer to downstream wireless perception applications.
• Run comprehensive experimental evaluations and comparisons with state-of-the-art approaches.
• Build a proof-of-concept demonstration of the proposed technology.
• Summarize research findings and contribute to a research publication and/or patent application.

Required profile

• Currently enrolled in a PhD program in Computer Science, Electrical Engineering, Computer Engineering, Robotics or a related field.
• Strong background in machine learning, deep learning or representation learning.
• Experience with wireless sensing or RF signal data, including channel measurements or related time-series signals.
• 2+ years of programming experience in Python and strong proficiency with PyTorch.
• Ability to independently design, implement and evaluate research ideas, and excellent written and verbal communication skills.

Required skills

• Preferred: experience with self-supervised learning, foundation models, transformers, contrastive learning or generative models.
• Preferred: experience with wireless sensing for presence detection, activity recognition, localization or smart-home sensing.
• Preferred: experience with RF data-acquisition pipelines, signal preprocessing and time-series feature extraction.
• Preferred: experience with multimodal learning and large-scale or distributed model training on GPUs.
• Preferred: publications in venues such as ICLR, NeurIPS, ICML, SenSys, IPSN, MobiSys, MobiCom, IMWUT/UbiComp or IROS.