Senior Research Scientist | Model Scaling

📋 Description

• Drive evaluation of open foundation/open-weight models for translation.
• Lead scaling decisions for hundreds of billions of parameters (MoE).
• Design multi-capability adaptation using LoRA and PEFT.
• Own modelling lifecycle: prototyping, ablations, scaling experiments, evaluation.
• Collaborate with post-training, RL/RLHF, and instruction teams to align the base model.
• Stay ahead of open-model literature and share recommendations.

🎯 Requirements

• Hands-on experience adapting and scaling LLMs via fine-tuning or post-training.
• Strong judgement on architecture trade-offs at scale (dense vs MoE).
• Knowledge of LoRA/PEFT and related methods.
• Hands-on training, experiments, and production deployment.
• Strong coding and experimentation skills (Python, PyTorch/JAX/TensorFlow).
• Clear communication and collaboration across teams.

🎁 Benefits

• Diverse international team across UK, Germany, Netherlands, Poland, US, Japan.
• Open communication and regular feedback culture.
• Hybrid work with office twice a week and flexible hours.
• Virtual Shares: ownership in DeepL's growth.
• Regular in-person team events and monthly hack Fridays.
• 30 days annual leave and mental health resources.