Principal Applied Scientist, Devices-Emerging Products, Amazon Devices - Emerging Products
Amazon Devices is building products that enable new forms of ambient computing, and we are seeking a Principal Applied Scientist to set the scientific direction for what comes next.
This position is based out of our SFO13 office in San Francisco, CA.
In this role, you will be the technical authority for bringing large-scale, multimodal AI to Amazon devices and services. You will work at the intersection of LLMs, VLMs, and multimodal perception, taking the latest research and making it work across cloud and edge environments, including compute-constrained consumer hardware. You will own the hardest, most ambiguous problems end to end: from large-scale data collection and model training to real-time on-device inference and the evaluation frameworks that tell us whether we are winning.
This is a hands-on, individual-contributor leadership role. You will set the scientific vision across multiple teams, raise the bar for scientific excellence, and influence product direction for experiences that will reach millions of customers. You thrive in ambiguity, move fast, and bring a critical eye toward what actually ships.
Key job responsibilities
* Define and drive the long-term scientific vision for multimodal AI across the organization, spanning perception, conversation, memory, and on-device intelligence
* Serve as the primary technical authority on multimodal AI methods (LLMs, VLMs, and multimodal perception), and advise technical and product leadership on the research bets worth making
* Identify and tackle intrinsically hard, open-ended research problems, acquiring expertise as needed and proposing innovative solutions that span multiple teams
* Drive the full cycle from research to production: large-scale real and synthetic data collection, model training and fine-tuning, optimization, and real-time deployment
* Develop and apply model compression, distillation, and optimization techniques to run large multimodal models on edge and compute-constrained devices, complemented by cloud-side pipelines
* Define rigorous evaluation frameworks, benchmarks, and gap-analysis cadences that align model quality with customer experience and business objectives
* Lead rapid experimentation, from paper to prototype to production decision, and deliver compelling demonstrations to senior leadership and partners
* Collaborate across science, engineering, hardware, design, and product to ship real-time, robust experiences within latency and compute constraints
* Mentor and develop a community of applied scientists and engineers, raising the scientific bar across the organization and setting the standard for excellence in artifacts, code, and methodology
About the team
Our team is part of Amazon Devices, Emerging Products, specializing in bringing advanced AI capabilities to Amazon devices. We operate with the cadence and structure of a fast-paced startup, with rapid iteration, high ownership, and plenty of opportunities for growth. If you want to work at the intersection of advanced research and real-world product impact, this is the place.
This position is based out of our SFO13 office in San Francisco, CA.
In this role, you will be the technical authority for bringing large-scale, multimodal AI to Amazon devices and services. You will work at the intersection of LLMs, VLMs, and multimodal perception, taking the latest research and making it work across cloud and edge environments, including compute-constrained consumer hardware. You will own the hardest, most ambiguous problems end to end: from large-scale data collection and model training to real-time on-device inference and the evaluation frameworks that tell us whether we are winning.
This is a hands-on, individual-contributor leadership role. You will set the scientific vision across multiple teams, raise the bar for scientific excellence, and influence product direction for experiences that will reach millions of customers. You thrive in ambiguity, move fast, and bring a critical eye toward what actually ships.
Key job responsibilities
* Define and drive the long-term scientific vision for multimodal AI across the organization, spanning perception, conversation, memory, and on-device intelligence
* Serve as the primary technical authority on multimodal AI methods (LLMs, VLMs, and multimodal perception), and advise technical and product leadership on the research bets worth making
* Identify and tackle intrinsically hard, open-ended research problems, acquiring expertise as needed and proposing innovative solutions that span multiple teams
* Drive the full cycle from research to production: large-scale real and synthetic data collection, model training and fine-tuning, optimization, and real-time deployment
* Develop and apply model compression, distillation, and optimization techniques to run large multimodal models on edge and compute-constrained devices, complemented by cloud-side pipelines
* Define rigorous evaluation frameworks, benchmarks, and gap-analysis cadences that align model quality with customer experience and business objectives
* Lead rapid experimentation, from paper to prototype to production decision, and deliver compelling demonstrations to senior leadership and partners
* Collaborate across science, engineering, hardware, design, and product to ship real-time, robust experiences within latency and compute constraints
* Mentor and develop a community of applied scientists and engineers, raising the scientific bar across the organization and setting the standard for excellence in artifacts, code, and methodology
About the team
Our team is part of Amazon Devices, Emerging Products, specializing in bringing advanced AI capabilities to Amazon devices. We operate with the cadence and structure of a fast-paced startup, with rapid iteration, high ownership, and plenty of opportunities for growth. If you want to work at the intersection of advanced research and real-world product impact, this is the place.