Overview As a Senior Applied Scientist on the Delivery Foundation Model team, you will advance foundation-model innovation to support Amazon's Last Mile delivery at scale. You will lead the development of multimodal architectures and robust production-ready solutions, working with world-class scientists and engineers. You'll tackle real-world data challenges across image, video, geospatial modalities and leverage massive infrastructure to push performance and reliability. This role offers influential research leadership with tangible operational impact.
Compensation / Benefits- Health insurance (medical, dental, vision)
- RSUs and sign-on payments
- 401(k) matching
- Paid time off and parental leave
- Flexible spending accounts
- Life insurance and disability coverage
Responsibilities- Design and implement novel multimodal foundation-model architectures (image, video, geospatial)
- Train and infer foundation models at Amazon scale using latest hardware and libraries
- Collaborate with cross-functional science and engineering teams to adapt models for Last Mile use cases
- Guide technical direction for research initiatives and ensure production robustness
- Mentor scientists while contributing hands-on technical work
- Develop and deploy scalable infrastructure for training, evaluation, and inference
- Drive focused technical initiatives from conception to production deployment
- Lead experiments, prototype new ideas, and support production integration
Key requirements- PhD in a quantitative field or a Master's with 10+ years of research/industry experience
- 5+ years building machine learning models or algorithms for business applications
- Proficient with data, SQL, and Spark
- Expert in production-level coding with Python and/or C++
- Strong publication record at top conferences OR demonstrated industry ML impact
- Experience mentoring junior scientists or engineers
- Mentoring and leadership
- Cross-team collaboration
- Strong communication of complex ideas
- Multimodal model architectures
- Foundation models development and adaptation
- Large-scale ML training and inference
