Senior AI Translation Model Architect

The Consensus; Greater London; Hybrid; GBP 110,000 - 140,000; Full time

Benefits offered by this job

Hybrid work schedule

Virtual Shares

Regular in-person team events

Hack Fridays

30 days annual leave

Job summary

DeepL is seeking a Senior Research Scientist to own the foundational modelling decisions for our next-generation translation models. This hands-on role includes prototyping, running large-scale experiments, and guiding modelling choices into production.

You will select foundation models, scale architectures to billions of parameters, and collaborate with post-training, RLHF, and instruction-following experts to integrate alignment and capability work into the base model.

Qualifications

• Proven ability to scale large language models.
• Experience prototyping and evaluating ML models at scale.
• Strong collaboration with cross-functional teams.

Responsibilities

• Drive foundation model selection and architectural decisions for translation systems.
• Prototype, ablations, and manage experiments at scale.
• Deliver research results into production with reproducible evaluation.
• Collaborate with post-training, RLHF, and instruction-following experts.

Skills

LLM scaling

Fine-tuning

Mixture-of-Experts

LoRA/PEFT

Experimentation

Python

PyTorch

JAX

TensorFlow

Production

Tools

Python

PyTorch

JAX

TensorFlow

Job description

DeepL is seeking a Senior Research Scientist to own the foundational modelling decisions for our next-generation translation models. This hands-on role includes prototyping, running large-scale experiments, and guiding modelling choices into production.

You will select foundation models, scale architectures to billions of parameters, and collaborate with post-training, RLHF, and instruction-following experts to integrate alignment and capability work into the base model.