Staff AI Research Engineer

Duolingo Duolingo · Consumer · Pittsburgh, PA · Engineering

Staff AI Research Engineer at Duolingo focused on the Monetization team, developing and deploying AI systems including bandit models and large language models to improve the learning experience and drive business objectives. The role involves full-stack ML development from feature engineering and training data to fine-tuning, reinforcement learning, evaluation, deployment, and monitoring.

What you'd actually do

  1. Contribute to the development and training of bandit models and large-scale neural network models.
  2. Collaborate with cross-functional teams to understand their needs, to align the models’ outputs with company objectives.
  3. Participate in and help drive strategic product and business decision making as a member of the Monetization leadership group.
  4. Stay up-to-date with the latest developments in machine learning, particularly in LLMs and bandits, and apply this knowledge to drive advancements in our projects.
  5. Ensure the delivery of high-quality, scalable, and efficient machine learning solutions.

Skills

Required

  • LLMs
  • multi-armed bandits
  • large-scale neural network models
  • feature engineering
  • training data development
  • fine-tuning
  • reinforcement learning
  • quality evaluations
  • deployment
  • monitoring
  • machine learning concepts
  • frameworks
  • best practices

Nice to have

  • computer vision

What the JD emphasized

  • proven track record as an AI research engineer
  • extensive experience across various machine learning techniques including large language models, multi-armed bandits
  • experience at all levels of the ML stack, including feature engineering, developing training data, fine-tuning, reinforcement learning, quality evaluations, deployment, and monitoring
  • broad experience in AI/ML is highly desirable
  • Proven experience as an AI research engineer, for example, in LLMs, bandits, computer vision, or related fields.
  • Strong background in training and fine-tuning large models in an applied setting.

Other signals

  • bandit models
  • large-scale neural network models
  • LLMs
  • fine-tuning
  • reinforcement learning
  • quality evaluations
  • deployment
  • monitoring