Principal Scientist, Natural Language Processing

3+ months agoBerlin, Germany


As a Principal Scientist on this growing team, you will take on a leading role in the Natural Language Processing (NLP) teams for Amazon Search. We are developing NLP components that cover a wide range of languages, from English and other Indo-European languages to Turkish, Arabic, Japanese and many more, and play a central role in search query processing, product description indexing, string normalization and vector embeddings for queries and products.

You will guide team members on the use of machine learning, tuned on terabytes of product and traffic data, to build or improve NLP technology that we integrate with the production search engine and evaluate using techniques like A/B tests. You will work with the Amazon Search leadership to set direction for the NLP approaches, model architectures and training methods, balancing business-defined performance indicators with the needs of millisecond response times. You will propose and explore publication-worthy innovation.

You will build relationships with stakeholders and partner teams across multiple countries, analyze data for trends, track down the source and meaning of anomalies, select suitable rule-based or machine-learning based techniques and advise team members as they make improvements, closing the loop through data, model, application and customer feedback.

• Define standards for the analysis of customer-focused problems in an industrial setting
• Guide team members and the leadership team on accurate, efficient and innovative NLP/ML solutions
• Get hands-on when building NLP models that can be applied to multiple languages
• Publish the developed innovative solutions in leading academic scientific venues in NLP/ML


Basic Qualifications
• PhD degree or equivalent, in a quantitative field (computer science, mathematics, natural language processing, artificial intelligence, or similar)
• At least 3 recent publications in a leading journal or conference related to computer science, natural language processing or applied mathematics
• 10+ years experience using machine learning or other NLP techniques, including but not limited to Deep Learning
• 4+ years experience implementing machine learning related techniques using Python or Scala in a distributed system, suitable for large scale
• Experience coaching or mentoring team mates (not necessarily in a manager role)
• Fluency in written and spoken English (German is not required)


Preferred Qualifications
• Experience working in data or applied science in a consumer product company
• Experience using Java or C++ writing production-ready code
• Familiarity with Deep Learning and machine learning frameworks such as SageMaker, PyTorch or TensorFlow
• Experience in low-level optimization of large-scale neural networks
• Fluency in one or more languages other than English

Job ID: Amazon-1088214