Deep Learning Research Engineer
Berlin, Hybrid
Full time
We are supporting a small research focused deep technology company that is developing domain specific foundation models for engineering applications.
The team works at the intersection of deep learning research and industrial product development. Its goal is to create new model architectures that can generalise across physical, geometric, and engineering problem classes.
This is an opportunity for an early career Research Engineer who wants to implement research ideas, run structured experiments, and gradually take ownership of original research questions.
The position
You will work closely with an experienced research team to turn mathematical and machine learning ideas into tested models and working systems.
Your responsibilities will include:
- Implementing, training, and evaluating new deep learning architectures.
- Reproducing published research and adapting methods to new engineering problems.
- Building reliable training, evaluation, and experiment pipelines.
- Generating and working with synthetic data.
- Running controlled experiments, benchmarks, and ablation studies.
- Investigating why existing approaches fail and proposing improvements.
- Documenting findings and communicating results clearly.
- Taking increasing ownership of research questions as you develop within the role.
Your background
You should have:
- A Master’s degree in computer science, mathematics, physics, engineering, or a closely related subject. An excellent Bachelor’s degree combined with strong practical or research experience may also be considered.
- Strong foundations in linear algebra, probability, optimisation, and deep learning.
- Experience independently implementing and training deep learning models.
- Strong Python programming ability.
- Experience with PyTorch, JAX, or TensorFlow.
- The ability to read research papers and translate ideas into working code.
- Clear written and spoken English.
Experience in one or more of the following areas would be useful:
- Geometric deep learning.
- Graph neural networks.
- Equivariant neural networks.
- Neural operators.
- Scientific machine learning.
- Physics informed machine learning.
- Surrogate modelling.
- Generative modelling.
- Reinforcement learning.
- Synthetic data generation.
- Rust, C++, CUDA, GPU computing, or high performance computing.
- Experiment tracking and GPU cluster environments.
- Research publications or substantial research projects.
Why consider this position?
You will join a small team where your work will directly influence the research direction and technical foundations of the product.
You will have broad ownership, close access to experienced researchers, and the opportunity to work on machine learning problems connected to physical and industrial systems.
The working model is hybrid in Berlin, with flexibility to work from home for part of the week.