Research Engineer, Robot Foundation Models
Vor Ort · Zurich (Schweiz) · Mid · Vollzeit
Mehr Stellen wie diese?
Erstelle dein kostenloses Profil bei PyHires: Du siehst, wie gut du zu jeder Stelle passt, und bekommst jede Woche eine E-Mail mit den Stellen, die zu dir passen.
Überblick
- Arbeitsmodell
- Vor Ort
- Standort
- Zurich (Schweiz)
- Anstellung
- Vollzeit
- Erfahrung
- Mid
- Sprachen
- Englisch
- Offen für Bewerber/innen aus
- Ganz Europa (EU und Nicht-EU)
- Unternehmen
- Nomagic
- Veröffentlicht
- 19. Nov. 2026
Über die Stelle
Nomagic has robots in production at scale and is using their massive deployment logs, the "internet data" of robotics, to train foundation models that move its fleet from a classical control stack to general physical AI. As a Research Engineer you will build, train and deploy those models, bridging ML research and industrial-scale robotic execution, with real robots every day.
What you'll do
- Own the training stack: design and maintain the infrastructure for large-scale VLA model training, including scheduling, distribution, job management, checkpointing and logging
- Build the tools to launch, monitor, debug and reproduce multi-variant experiments
- Pre-train robot foundation models from the repository of offline, classical-stack deployment data
- Turn research needs into infrastructure capabilities, track experiments and close the loop with ML researchers
- Design physical benchmarks and run structured on-robot evaluations to measure real-world success rates
- Analyse evaluation results to guide research, and document workflows so the team can scale them
What you bring
- Experience training, fine-tuning and deploying modern deep learning architectures (Transformers, VLMs or VLAs, imitation learning, RL) for robot control, ideally validated on real hardware
- High proficiency in Python and deep learning frameworks (PyTorch/JAX), with clean, scalable code for training and evaluation
- Robotics intuition: comfort with hardware and an understanding of perception, controls and state estimation
What they offer
- Real robots and real problems every day
- A relocation package and flexible working hours
- An English-speaking environment and a focused team where your work has direct impact
- A final decision within 2-3 days after the on-site interview, with detailed feedback
Anforderungen
Basiert auf:Python
Muss
- JAX
- PyTorch
Über Nomagic
Nomagic builds AI for industrial robots, with robots already in production at scale in real factory and warehouse deployments. It is now working on general-purpose physical AI, using the deployment logs of its fleet as training data for robot foundation models.
Stack
- Python
- JAX
- PyTorch
Unternehmen
Nomagic
Schweiz