Senior Research Scientist | Multimodal Systems

As Senior Research Scientist at DeepL, you will lead fine-tuning, post-training, and reinforcement learning for next-generation document translation multimodal and vision models. You’ll own a key research direction, prototype rapidly, and scale experiments into production. You’ll advance models that reason about document layout by fusing real-world and synthetic data, making translations more steerable and adaptable. This role sits within our collaborative Language AI team, driving impact from research to real-time systems.

• Drive development of vision and multimodal models for document, image and media translation across ingestion to end-to-end systems
• Lead post-training research and development for vision/multimodal models: supervised fine-tuning, knowledge distillation, preference optimization, and RL tuned to translation quality
• Build evaluator models for document and design quality, including rubric- and reference-based grading; investigate reward hacking and quality-estimation failures
• Own full model delivery lifecycle: prototyping, ablations, training, evaluation, optimization, and production deployment with engineering to ship at scale
• Establish evaluation, reproducibility, monitoring, and continuous improvement practices for production models

• Proven experience with developing multimodal models, VLM, or vision models
• Deep hands-on expertise in model post-training, knowledge distillation, and reinforcement learning (RLHF/RLAIF, PPO/GSPO, reward modeling)
• Strong data-centric instincts for synthetic-data and preference-data pipelines, data curation and filtering, augmentations, and data mixtures/ablations
• Hands-on builder who trains models, runs experiments, debugs pipelines, and ships ML systems in production with a product impact focus
• Strong coding and experimentation skills (Python, PyTorch/JAX/TensorFlow) and ability to align research with product and engineering priorities
• Ability to lead complex research efforts and collaborate across teams with focus on real-world performance and user experience

• collaboration
• clear communication
• problem-solving
• Multimodal models
• Vision models
• Post-training techniques