London Hybrid Full-time

DeepL is hiring a Senior Research Scientist | Model Steering

Responsibilities

  • Lead the creation of translation models that can be guided by user-defined preferences, contextual signals, and rule-based inputs.
  • Conduct hands-on research into post-training methods for core translation systems, including supervised fine-tuning, knowledge distillation, preference alignment, and reinforcement learning for quality optimization.
  • Develop reward and evaluation models for translation performance, using rubric-based and reference-based scoring, while identifying and addressing reward manipulation and quality estimation failures.
  • Advance research toward models that process multimodal inputs and contextual data to enhance translation accuracy and relevance.
  • Manage end-to-end model development, from prototyping and ablation studies to training, evaluation, optimization, and deployment in scalable real-time systems, in close collaboration with engineering teams.
  • Implement robust standards for model evaluation, reproducibility, production monitoring, and iterative improvement.
  • Guide and mentor researchers and engineers, fostering a collaborative environment focused on raising model quality and technical excellence.

Benefits

  • Global, multicultural team with representation from over 90 nationalities, operating across multiple countries and time zones.
  • Culture of transparent communication, constructive feedback, and empathetic collaboration grounded in a growth mindset.
  • Hybrid work model with office presence two days per week, balancing in-person interaction with remote flexibility.
  • Virtual Shares program that grants every employee a stake in company performance and long-term success.
  • Frequent in-person gatherings, including team meetups, onboarding events, and company-wide conferences.
  • Monthly dedicated innovation days allowing team members to explore passion projects and cross-team collaboration.
  • Thirty days of paid annual leave per year, in addition to public holidays.
  • Comprehensive benefits package tailored to local regulations and employee needs across global locations.

Compensation

Competitive benefits

Work Arrangement

Hybrid

Team

Diverse and internationally distributed team

Other

  • 30 days of annual leave (excluding public holidays)
  • Access to mental health resources
  • Flexible working hours
  • Hybrid work schedule with office attendance twice a week

Not specified

About company
DeepL
A global communications platform powered by Language AI that provides translations and intelligent writing suggestions for over 100,000 businesses worldwide.
All jobs at DeepL Visit website
Job Details
Department Research
Category Data & ML
Posted 8 days ago