Research Engineer - Large-Scale Training (Remote + Equity)

Black Forest Labs; San Francisco (CA); Hybrid; USD 180,000 - 290,000; Full time

Job summary

Black Forest Labs is seeking an expert specialized in training systems to enhance performance and stability for multimodal generative models, concentrating on GPU-level optimizations and distributed systems. The role involves direct collaboration with researchers to implement efficient training solutions.

Candidates should possess strong skills in PyTorch, experience with large-scale training, and practical judgment for profiling GPU workloads. Hybrid working options are available, alongside a compensation package that includes a competitive salary range.

Qualifications

• Experience improving training throughput, memory footprint, or stability.
• Strong judgment in profiling GPU workloads.
• Familiarity with low-precision training and quantization.

Responsibilities

• Improve production training performance for multimodal generative models.
• Profile training steps for various components.
• Debug distributed training failures.

Skills

PyTorch fluency

Large-scale training systems

Distributed training concepts

GPU workload profiling

Low-precision training understanding

Job description

Black Forest Labs is seeking an expert specialized in training systems to enhance performance and stability for multimodal generative models, concentrating on GPU-level optimizations and distributed systems. The role involves direct collaboration with researchers to implement efficient training solutions.

Candidates should possess strong skills in PyTorch, experience with large-scale training, and practical judgment for profiling GPU workloads. Hybrid working options are available, alongside a compensation package that includes a competitive salary range.