Distributed Training Engineer (Remote)

About the role

As a Distributed Training Engineer you will lead the scaling and optimization of large‑scale machine‑learning workloads across multi‑GPU and multi‑node environments. You will work remotely with a global team of top‑tier engineers, ensuring that massive datasets and models run efficiently and reliably.

Key responsibilities

• Implement and fine‑tune distributed training strategies using PyTorch DDP, FSDP, and DeepSpeed.
• Profile training performance and GPU utilization with Nvidia Nsight and related tools.
• Design mixed‑precision and memory‑optimization pipelines to maximize throughput and reduce footprint.
• Resolve coordination and synchronization challenges across multi‑node clusters using CUDA and Triton.
• Scale large models through parameter sharding, pipeline parallelism, and efficient data sharding techniques.
• Collaborate with ML research teams to align infrastructure with evolving model requirements.

Required profile

• 4+ years of experience in Machine Learning Engineering or Distributed Systems.
• Proven expertise with distributed training frameworks such as PyTorch DDP, FSDP, or DeepSpeed.
• Hands‑on experience with Nvidia GPU profiling tools (e.g., Nsight).
• Strong proficiency in CUDA programming and multi‑node performance optimization.
• Experience with inference optimization tools like Triton or TensorRT.
• Advanced proficiency in English.

Required skills

• PyTorch DDP
• FSDP
• DeepSpeed
• Nvidia Nsight
• CUDA
• Triton
• TensorRT
• Mixed‑precision training
• Parameter sharding
• Pipeline parallelism
• Data sharding

What we offer

• 100% remote work from anywhere.
• Competitive compensation in USD or local currency.
• Provided hardware and software setup for home office.
• Flexible hours and self‑managed schedule.
• Paid parental leave, vacations, and national holidays.
• Mentorship, promotion pathways, and continuous skill development.