About the Role
Join an AI research team to turn experimental code into reliable, production-ready systems. You will build post-training pipelines for generating, processing, and evaluating research data, helping research ideas scale into repeatable experiments.
What You'll Do
Convert research prototypes into tested, reusable data generation, training, and evaluation pipelines.
Build distributed experiment support for data loading, checkpointing, and collecting agent interactions.
Profile workloads and improve GPU utilization, memory efficiency, and data throughput.
Develop tests, experiment tracking, and debugging tools while preserving the integrity of experimental results.
Collaborate with researchers to turn ideas into robust, maintainable software.
What We're Looking For
Typically 2 to 10 or more years of relevant experience building machine learning training, inference, or data pipelines in Python.
Strong Python skills and hands-on experience with PyTorch, JAX, or a comparable machine learning framework.
Experience building and debugging software for model training, inference, or data processing at scale.
Understanding of machine learning experiments, including how data, numerical precision, and implementation choices affect results.
Sound software engineering practices, including testing, profiling, version control, and documentation.
Experience with distributed tools such as PyTorch distributed, DeepSpeed, or Ray is helpful, as is familiarity with Transformers libraries, vLLM, or experiment tracking tools.
Background in reinforcement learning systems or working with video, audio, or image datasets is a plus.
Compensation & Benefits
Competitive compensation with equity is offered, along with health, dental, and vision coverage and professional development support. Visa sponsorship and relocation support are available.
Location
This is a full-time, on-site role in Zürich, Switzerland.