Machine Learning Engineer

About the role

MBN Solutions is seeking a researcher to help develop a new class of foundation models focused on cause and effect rather than standard language processing. You will work on an early-stage team of fewer than 15 people to build models that learn from physical systems and real-world observational data.

Key facts

What you'll do

• Build and train large-scale foundation models from the ground up.
• Create and test original model architectures.
• Manage distributed training across hundreds of GPUs.
• Construct data pipelines capable of handling petabyte-scale multimodal information.
• Build frameworks for model evaluation and interpret experimental data.
• Partner with researchers to test new concepts and refine the core research agenda.

Requirements

• Experience training foundation models from scratch instead of fine-tuning.
• Proficiency in distributed training using PyTorch, FSDP, or comparable frameworks.
• Background in developing large-scale machine learning infrastructure and data pipelines.
• Strong grasp of machine learning fundamentals.
• Ability to work across both research and engineering tasks.

Nice to have

• Experience with Large Language Models, Computer Vision, or Robotics.
• Knowledge of Sensor Fusion, Physics-informed Machine Learning, Scientific AI, or Multimodal Learning.

Skills & tools

• PyTorch
• FSDP
• Distributed training systems
• Large-scale data pipelines
• Multimodal foundation models

Practical notes

• Relocation support is available for this role.