Amazon Applied Scientist III Prime Video - Compression Efficiency Research Team 

United States, Washington, Seattle 

557241113

Prime Video offers customers a vast collection of movies, series, and sports—all available to watch on hundreds of compatible devices. U.S. Prime members can also subscribe to 100+ channels including Max, discovery+, Paramount+ with SHOWTIME, BET+, MGM+, ViX+, PBS KIDS, NBA League Pass, MLB.TV, and STARZ with no extra apps to download, and no cable required. Prime Video is just one of the savings, convenience, and entertainment benefits included in a Prime membership. More than 200 million Prime members around the world enjoy access to Amazon’s enormous selection, exceptional value, and fast delivery.As a Prime Video technologist, you’ll have end-to-end ownership of the product, user experience, design, and technology required to deliver state-of-the-art experiences for our customers. You’ll get to work on projects that are fast-paced, challenging, and varied. You’ll also be able to experiment with new possibilities, take risks, and collaborate with remarkable people.This position involves research into and development of advanced DL models for video assets in order to improve their visual quality to be appealing to our customers (e.g. restoration from various defects, improving resolution, restoring truer colors, SDR->HDR up-conversion, better dithering/noising techniques to withstand compression artifacts, etc.). You will also develop next-generation methods for content-adaptive video encoding. You will work closely with subject matter experts (Research Scientists) with depth in the video encoding/quality/processing areas to maximize the video quality that Prime Video customers experience under a given network condition.Key job responsibilities
The work would involve working with creation of ground-truth datasets, creating new CNN, transformer and visual foundation model architectures and training methodologies that ensure the best temporal consistency and perceptual quality, optimization and deployment of such models in production workflow, and defining/measuring suitable at-scale success criteria.Aspects of work can involve scene understanding at a semantic level using latest techniques. You would develop innovative solutions that achieve the best balance between speed vs performance, file invention disclosures, and also publish these novel approaches in tier-1 conferences . You would collaborate with research scientists with domain depth in video encoding/compression. You would conduct subjective ratings tests to generate ground truth data for your models, when required. You would also contribute to development of new video quality models that correlate highly with subjective ratings and help quantify the improvements brought about by your work. You would define and refine team processes around ML/DL model development, deployment, and guardrail definition.A day in the life
You will work with Video Research Scientists who will bring domain expertise to jointly define the specific problem and technical approach. You may collaborate with academia through sponsored research programs on complex problems of long-term importance in your area of work. You will interface with ML engineers to convert your output into production workflows. You will help develop new AI workflows that are more efficient from a compute perspective. You will work with product managers to develop product roadmap that align with business goals. You will help hire new talent and will help develop talent through knowledge sharing and mentorship.

- 3+ years of building machine learning models for business application experience
- PhD, or Master's degree and 6+ years of applied research experience
- Experience programming in Java, C++, Python or related language
- Experience with neural deep learning methods and machine learning
- Experience in patents or publications at top-tier peer-reviewed conferences or journals