Tower Research Capital
Machine Learning Performance Engineer, Training
$200K
New York, NY +1 · Full-time · Posted Sep 24, 2026
- Location
- New York, NY
New York, NY - 120 Broadway - Experience
- 3+ years optimizing machine-learning training workloads in high-performance, distributed, or large-scale computing environments.
About the job
Build and optimize scalable machine-learning training systems, improving performance from data ingestion and distributed execution through GPU kernels and infrastructure utilization.
What you'll do
- Benchmark training workloads and develop performance models to guide compute infrastructure decisions.
- Optimize distributed training, GPU kernels, and end-to-end training pipelines to improve throughput and accelerator utilization.
- Collaborate with research, HPC, systems, and hardware teams to build efficient, reliable training infrastructure.
What they ask for
- Deep knowledge of PyTorch or JAX, including execution models, compilation, autograd, and distributed training.
- Strong Python and C++ programming skills, including performance-critical systems.
- Experience developing or optimizing GPU kernels, and understanding of GPU architecture.
- Experience with distributed training, performance profiling, and benchmarking heterogeneous compute platforms.
Nice to have
- Experience optimizing computationally intensive models such as transformers, time-series, or reinforcement learning.
- Experience with cluster orchestration or scheduling technologies such as Kubernetes, Slurm, or Ray.
Pay
$200,000 USD annual base salary
Tools
PyTorch, JAX, Python, C++, CUDA, Triton, NCCL, Nsight Systems
AI focus
Accelerate large-scale machine-learning model training through distributed training optimization, GPU kernel development, numerical and model optimization, and improved training infrastructure.
Model-written summary · the original posting is authoritative