PyTorch Internals Expert — Project Torch

We're sourcing ML, systems, and distributed systems engineers with deep, internals-level PyTorch experience (e.g. custom autograd, CUDA/tensor extensions, ATen-level work, distributed training internals) — full-time professional or research experience required. Complete up to 3 short PyTorch tasks (~2 hrs active work each, plus background runtime). Pay: $150/task. Fully remote, rolling onboarding next week.

Responsibilities

• Complete up to 3 assigned technical tasks involving low-level PyTorch work
• Ensure code runs correctly through full runtime, including any extended, unattended runtime
• Complete tasks independently within a rolling onboarding schedule

Requirements

• Full-time professional or research experience with PyTorch
• Demonstrated experience with internals-level PyTorch work — e.g. custom autograd functions, tensor/CUDA extensions, or ATen-level work
• Access to suitable hardware (GPU-enabled machine or cloud instance)

Preferred Qualifications

• Background in ML, systems engineering, or distributed systems engineering
• Experience with distributed training internals
• Experience with compiler/graph-level work or numerical/algorithmic runtime optimization
• Contributions to PyTorch or adjacent open-source libraries

Why Apply

• Most tasks require only ~2 hours of active hands-on work
• $150 per task, up to 3 tasks (with potential for a bonus)
• Fully remote