MLDL - Bilingual, Bedrock

Amazon - Boston, MA

Full-time

Boston, MA

Sporting Goods, Hobby, Musical Instrument, Book, and Miscellaneous Retailers

About the position

The Bedrock team is focused on supporting the training of various models in the AWS generative AI platform. This role involves working with different model types, generating data for machine learning model training, and evaluating toxic content. The position emphasizes collaboration among data linguists to ensure high-quality data annotation and the application of responsible AI practices.

Responsibilities

• Build a thorough understanding of data collection and annotation guidelines and various annotation tools.
• Annotate, generate and QA data, identifying linguistic categories based on detailed annotation and adhering to guidelines.
• Use generative AI to facilitate workflows or automate repetitive tasks.
• Monitor AI outputs for biases or ethical issues and adjust inputs to mitigate these risks.
• Perform annotation related tasks; participate in data generation, collection and quality assurance tasks.
• Collaborate with other ML Data Linguists to resolve data ambiguities and annotation disagreements.
• Perform qualitative error trend analysis and devise action plans to improve data quality.
• Provide feedback to Language Engineers and Scientists on tool improvements and annotation processes.
• Implement solutions independently for identified issues.
• Contribute to process improvements to reduce handling time and improve resource output.
• Develop language artifacts crucial for model development such as datasets for training and evaluation.
• Support and consult in pre-screening interviews for Data Associates.
• Collaborate with LEs, scientists, and Ops Manager to innovate processes, tracker automations, and workflows.
• Assist LEs in communication with vendors to provide detailed feedback to annotators.

Requirements

• Bachelor's degree in Linguistics, Philosophy, Cognitive Science, a foreign language, or Literature.
• Ability to identify linguistic ambiguity and inaccuracies in linguistic data, as well as identify basic parts of speech and produce reports of analyzed data.
• At least 6 months of experience with natural language data labeling, data annotation, linguistic annotation, or teaching experience, as well as experience leading a team of peers.
• Knowledge of different domains such as Finance, Health Care, and/or Insurance.
• Ability to generate innovative and diverse inputs to explore various aspects of an AI model's capabilities.
• Familiarity with json, yaml, xml or other forms of text markup.
• Ability to navigate a Unix terminal and use common command line tools.
• Knowledge of Python, Java or any other scripting language.
• Strong organizational and leadership skills and detail-oriented.
• Ability to communicate well and actively listen with other data associates on a team.
• Ability to deliver high quality results under tight deadlines.
• Comfortable working in a fast-paced, collaborative work environment.
• Willingness to support several projects at one time, and to accept re-prioritization as necessary.

Nice-to-haves

• Master's degree in a relevant field, such as Linguistics, Communications, a foreign language, computational linguistics or other language, data or tech related disciplines is a plus.
• Proficient in another foreign language.
• Familiarity with common text processing tools.
• Passion for language, linguistics, human language technology and AI.
• Ability to work in different operating systems (Windows, MacOS, or Linux).
• Strong understanding of NLP concepts and techniques.