Research Engineer / Scientist – Reinforcement Learning & LLM Post-Training

Next Ventures · Bremen, Germany
Location
Bremen, Germany
Work mode
Hybrid
Type
Full-time
Seniority
Senior
Salary
120,000 NON
Posted
19 Nov 2026

Verified open 19 Nov 2026 · Source: Company careers page

About the role

Research Engineer / Scientist – Reinforcement Learning & LLM Post-Training

Location: Europe (Remote / Hybrid)
Employment: Full-time, Permanent

The Opportunity

We are looking for an experienced Research Engineer or Scientist to help develop and improve the next generation of large language models.

This is an opportunity to take ownership of model training from initial experimentation through to large-scale reinforcement learning runs and evaluation. You will work across research and engineering, developing techniques that improve model capabilities, reliability and efficiency.

The focus is on training models capable of handling complex, multi-step tasks, interacting with external tools and performing reliably in demanding real-world environments.

You will join a technically strong, research-led team with access to substantial GPU infrastructure and the freedom to influence the direction of its work.

Key Responsibilities

Lead and execute end-to-end LLM post-training, including supervised fine-tuning and large-scale reinforcement learning.


Research and implement new RL algorithms, training strategies and optimisation techniques.


Develop training curricula, data strategies and reward mechanisms to improve model performance.


Investigate training behaviour, identify performance bottlenecks and improve training stability.


Optimise model efficiency, inference costs and performance across complex tasks.


Work with distributed training infrastructure and large-scale GPU clusters.


Develop reproducible experiments, evaluation frameworks and training pipelines.


Contribute to the development and release of open-source models and research outputs.

Experience & Technical Requirements

Practical experience training or fine-tuning large language models.


Strong understanding of reinforcement learning techniques, ideally including PPO, GRPO, RLHF or reinforcement learning with verifiable rewards.


Excellent Python and PyTorch skills.


Experience with distributed training frameworks such as FSDP, DeepSpeed or Megatron.


Familiarity with large-scale compute environments, including Slurm or Kubernetes.


Strong experimental methodology and the ability to interpret research results critically.


Comfortable working across both research and engineering challenges.


Ability to communicate technical findings clearly and work independently.

Desirable Experience

Experience with inference frameworks such as vLLM or SGLang.


Familiarity with RL frameworks such as VERL, OpenRLHF or TRL.


Experience debugging multi-node training environments and GPU infrastructure.


Knowledge of GPU optimisation and performance engineering.


Academic publications, research contributions or meaningful open-source projects.

Why Join?

Work on technically challenging problems at the forefront of AI research.


Take ownership of large-scale model training rather than working on isolated parts of the process.


Access to significant GPU compute and modern research infrastructure.


Opportunity to contribute to open-source AI development.


Join a small, ambitious research team where individual contributions have a meaningful impact.


Work closely with experienced technical leadership and influence the company's research direction.


Competitive compensation and potential equity participation.

Location & Working Arrangements

Flexible remote working arrangements across Europe.


Preference for candidates based within the EU.


Office access available in Germany for those who prefer a hybrid environment.


Occasional in-person attendance for team collaboration and company events.

Application Process

We are interested in understanding what candidates have actually built and the technical problems they have solved.

Applications should include a CV and examples of relevant work, such as research papers, GitHub repositories, technical projects or previous training experiments.

Candidates do not need to meet every listed requirement. Practical ability, research depth and a genuine interest in advancing AI capabilities are the most important considerations.

Similar jobs in Bremen at other employers

More jobs at Next Ventures

Browse similar jobs