Applied Scientist, Translation Services

DESCRIPTION

The Translation Services (TS) team is looking for a world-class Applied Scientist to join our India Science team and use AI to break down visual language barriers for Amazon customers worldwide. Our team is pioneering image translation technology that enables customers to understand product images, infographics, and visual content in their native language. As part of the team, we expect that you will develop innovative solutions to hard problems at massive scale, and publish your findings at peer-reviewed conferences and workshops.
With the recent advancements in Visual Language Models (VLM), Translation Services is reimagining how we make visual content accessible across languages. We're looking for Applied Scientists to deliver both the short-term roadmap working closely with Product and Engineering teams, and make strategic investments for the long-term. Our research themes include, but are not limited to: Visual Language Models, multimodal machine translation, scene text detection and recognition (OCR), contrastive learning, diffusion models for image-to-image translation, few-shot and zero-shot learning, transfer learning across languages and scripts, unsupervised and semi-supervised methods, active learning for multilingual visual understanding, and large-scale image translation systems. Key job responsibilities Innovate in visual translation: Design, develop, and deploy state-of-the-art machine learning models for image translation, including text detection, recognition, and translation in images across multiple languages and scripts. Advance VLM research: Collaborate with cross-functional teams of engineers, product managers, and scientists to identify and solve complex problems in Visual Language Model space applied to translation use cases. Experiment and iterate: Design and execute experiments to evaluate the performance of different models across diverse image types, languages, and domains, and iterate quickly to improve results. Build robust evaluation frameworks: Create comprehensive evaluation frameworks for assessing model performance across different languages, scripts (Latin, CJK, RTL, Indic), visual domains, and translation quality metrics. Think big and long-term: Envision the future of visual translation over a multi-year horizon, and identify new opportunities to apply VLM technologies to solve real-world translation problems at Amazon scale. Drive scientific excellence: Publish your research at top-tier conferences. Collaborate across teams: Work closely with Applied Scientists, MLEs, and engineers to translate research prototypes into production systems serving millions of customers. Communicate impact: Present results and insights to both technical and non-technical audiences, including through presentations, written reports, and scientific publications. About the team Translation Services, part of Amazon Exports and Expansions (AEE), powers Amazon's global reach through cutting-edge translation technology. In a recent year, 7,800+ users from 180+ Amazon teams used our systems to translate 274 billion words across retail, AWS, customer service, and more. Our India Science team is now extending this impact to the visual domain, making images and visual content accessible across languages at unprecedented scale. Our Vision: Language will not stand in the way of anyone on earth using Amazon products and services. Our Mission: We are the enablers and guardians of translation for Amazon's customers.

BASIC QUALIFICATIONS

• Experience programming in Java, C++, Python or related language
• Master's degree
• Experience researching about machine learning, deep learning, NLP, computer vision, data science

PREFERRED QUALIFICATIONS

• Experience in state-of-the-art deep learning models architecture design and deep learning training and optimization and model pruning
• Have publications at top-tier peer-reviewed conferences or journals