Senior AI Researcher - Pre-training (f/m/d)
Core
Designing and optimizing large-scale foundation model pre-training architectures and training recipes for high-stakes European applications.
Role type
Senior AI Researcher (Pre-training)
Builds
Next-generation foundation models for finance, manufacturing, and public administration
Domain
AI/ML, Foundation Models, Large-Scale Distributed Training
Deliverable
production ML models
Required skills
Python, PyTorch, Transformer training dynamics, Distributed training debugging, Mathematical reasoning, Experiment design, Software engineering, Model architecture design, Training stability analysis, Scaling strategy
Preferred skills
LLM/Multimodal model training, Distributed training frameworks (torchtitan, Megatron-LM, DeepSpeed), Scaling laws, Sparse training (MoE), Top-tier conference publications
Technologies
PyTorch, NVIDIA Blackwell GPUs, torchtitan, Megatron-LM, DeepSpeed
Responsibilities
Optimize training recipes (optimizers, learning rates, regularization), Develop scaling strategies and hyperparameter transfer methods, Design and evaluate model architectures, Diagnose convergence and stability issues, Collaborate on system-model co-design, Debug large-scale distributed training failures
Seniority
Senior, hands-on IC