Distributed Training and Inference Engineer

About the role

Sciforium is seeking a Distributed Training and Inference Engineer to build, optimize, and maintain the software stack that powers large‑scale AI training and serving workloads. You will work across low‑level CUDA/ROCm runtimes to high‑level frameworks such as JAX and PyTorch, ensuring performance, scalability and stability.

Key responsibilities

• Maintain and optimize critical ML libraries and frameworks (JAX, PyTorch, CUDA, ROCm) across diverse hardware configurations.
• Own the end‑to‑end ML stack from drivers to high‑level tooling, continuously improving performance.
• Optimize distributed training systems, ensuring efficient sharding, partitioning and configuration for large models.
• Integrate and validate modules for runtime correctness, memory efficiency and multi‑node scalability.
• Profile compilation graphs and training workloads, eliminating bottlenecks.
• Troubleshoot complex hardware‑software interactions, including kernel‑level issues and memory leaks.

Required profile

• 5+ years of industry experience in ML systems or distributed training.
• Bachelor’s or Master’s in Computer Science, Computer/Electrical Engineering or related field.
• Strong programming skills in Python and C++.
• Deep understanding of profiling tools such as Nsight, ROCm Profiler and XLA profiler.
• Experience with multi‑node distributed training frameworks (DTensor, GSPMD, etc.).

Required skills

• Python
• C++
• JAX
• PyTorch
• CUDA
• ROCm
• Nsight
• ROCm Profiler
• XLA Profiler
• DTensor
• GSPMD
• NCCL

What we offer

• Medical, dental and vision insurance.
• 401(k) retirement plan.
• Daily lunch, snacks and beverages.
• Flexible time off.
• Competitive salary and equity.
• Equal opportunity employment.