Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will analyze internal feature representations in multimodal perception foundation models and develop active-learning and data-curation algorithms for rare, safety-critical cases. You will explore vision-language-model data-labeling workflows, train and fine-tune neural networks, and validate performance using large-scale autonomous-vehicle sensor data and simulation.
Requirements
- Be currently enrolled in a PhD program in Computer Science, Robotics, Electrical Engineering, or a related quantitative field
- Program in Python and C++
- Have practical experience training, fine-tuning, and evaluating deep-learning models for computer vision or multimodal perception
- Understand modern neural architectures, including Vision Transformers and multimodal sensor encoders
Responsibilities
- Analyze and characterize internal feature representations of deep multimodal perception foundation models
- Design and implement active learning and intelligent data-curation algorithms to identify rare safety-critical edge cases
- Explore automated data-labeling and validation workflows using vision-language models
- Train, fine-tune, and evaluate deep neural networks
- Validate model performance on large-scale autonomous-vehicle sensor datasets and simulation environments
Benefits
- Medical, dental, and vision insurance
- Mental wellness support
- Gym membership and wellness programs
- Bonus opportunities
- Equity
- Employee discounts
- Flexibility to work from another location for four weeks per year
- Paid time off
- Bereavement, sick, and parental leave
- 26 weeks of paid leave for birthing parents
- 18 weeks of paid leave for non-birthing parents
- Education reimbursement
- Volunteer time off
- Access to Google offices, cafes, wellness centers, and massages
- At-home fitness and cooking classes