Overview
As Senior Research Scientist at DeepL, you will own foundational modelling decisions behind next-generation translation models. You will prototype rapidly, run large-scale experiments, and drive architecture choices for scaling open models, working closely with post-training, RL, and instruction-following experts. You’ll influence which foundation models to build on and how to adapt base models for production-ready translation systems. Join a globally distributed, innovation-driven team shaping the future of language AI.
Leistungen / Benefits
- Hybrid work
- Virtual Shares
- 30 days of annual leave
- Regular in-person team events
- Monthly hack Fridays
- Competitive location-aware benefits
Verantwortungsbereiche
- Drive selection and evaluation of open foundation/open-weight models for next-generation translation systems
- Lead model selection and architecture decisions to scale to hundreds of billions of parameters (including Mixture-of-Experts and sparse/efficient designs)
- Design multi-capability adaptation strategies using LoRA, PEFT, and related methods
- Own the modelling lifecycle: prototyping, ablations, scaling experiments, evaluation, and production delivery
- Collaborate with post-training, RL/RLHF, and instruction-following specialists to align base models with product goals
- Stay ahead of open-model and scaling literature and provide well-founded recommendations to the team
Zentrale Anforderungen
- Hands-on experience adapting and scaling large language models via fine-tuning/instruction-tuning/post-training of multi-billion-parameter models
- Sound judgment on architecture trade-offs at scale (dense vs. MoE) and choice of open-weight foundation models
- Knowledge of parameter-efficient and multi-capability adaptation (LoRA/PEFT) and variants
- Hands-on model training, experimentation, and debugging pipelines with production-ready results
- Strong coding and experimentation skills (Python, PyTorch/JAX/TensorFlow)
- Clear communication and cross-team collaboration to align research with product and engineering priorities
- Strong communication
- Collaborative mindset across teams
- Problem-solving and debugging mindset
- Python
- PyTorch
- JAX
Senior Research Scientist | Model Scaling in Bonn Arbeitgeber: DeepL
DeepL ist ein hervorragender Arbeitgeber, der eine dynamische und unterstützende Arbeitsumgebung bietet, in der Innovation und persönliches Wachstum gefördert werden. Mit flexiblen Arbeitszeiten und der Möglichkeit, remote zu arbeiten, ermöglicht DeepL seinen Mitarbeitern, ihre Work-Life-Balance zu optimieren, während sie an bedeutungsvollen Projekten im Bereich KI-Technologie arbeiten. Die Unternehmenskultur basiert auf offener Kommunikation und Teamzusammenhalt, was durch regelmäßige Teamevents und monatliche Hackdays unterstützt wird, um Kreativität und Zusammenarbeit zu fördern.
