MLOps Engineer - Hybrid, On-Prem & Edge Inference Lead

Transvalor Sa; Valbonne; Hybride; EUR 52 000 - 65 000; Plein temps

Avantages offerts par ce poste

Housing aid

Daily transport coverage

RTT (reduction of working time)

Occasional work from home

Résumé du poste

Physora, a Transvalor Group company, is seeking an MLOps Engineer to oversee the full lifecycle of AI surrogate models for material-forming processes.

You will build scalable training infrastructure, enable industrialization, and ensure reproducibility with in-house datasets, model registry, and deployment on-premise or at the edge. Collaboration with R&D and production teams is essential to deliver reliable, high-performance solutions.

Qualifications

• Master’s degree or PhD in Computer Science (specialization in Machine Learning / AI) or Computational Physics with significant experience in distributed training systems engineering.
• Substantial production experience with training infrastructure; not purely academic.
• Deep expertise in PyTorch and its ecosystem, including distributed training and profiling.
• Comfort with large volumes of numerical data and storage formats (HDF5, Zarr, Parquet).
• Strong software engineering discipline: testing and version control.
• Curiosity about physics and willingness to understand complex material-forming processes.

Responsabilités

• Build and maintain training infrastructure: loading large-scale results, distributed multi-GPU training, experiment tracking, model registry, reproducibility.
• Develop tooling to transform raw simulation outputs into usable datasets with storage schemas and versioning.
• Achieve required inference performance: quantization, export, serving within plant latency budgets (on-premise/edge).
• Run experiments and ablations; present results to inform architectural decisions.
• Contribute to model implementation alongside the modeling engineer.
• Bring production discipline from outside and anticipate past failures.

Connaissances

Distributed training systems

PyTorch ecosystem

High-performance computing

Software testing

Formation

Master’s/PhD in CS or Computational Physics

Outils

PyTorch

HDF5

Zarr

Parquet

MPI

C++

Description du poste

Physora, a Transvalor Group company, is seeking an MLOps Engineer to oversee the full lifecycle of AI surrogate models for material-forming processes.

You will build scalable training infrastructure, enable industrialization, and ensure reproducibility with in-house datasets, model registry, and deployment on-premise or at the edge. Collaboration with R&D and production teams is essential to deliver reliable, high-performance solutions.