Tower Research Capital
Machine Learning Performance Engineer (Inference)
$200K–$300K
New York +1 · Full-time · Posted Aug 10, 2026
- Location
- New York
New York, NY, United States - Experience
- 2+ years; any domain. Financial trading experience is not required.
About the job
Optimizes and deploys machine-learning inference pipelines for latency-critical trading systems, evaluating hardware and improving performance, efficiency, and reliability.
What you'll do
- Benchmark CPUs, GPUs, and FPGAs to inform infrastructure decisions and optimize execution across memory hierarchies and interconnects.
- Develop optimized GPU kernels and apply model-reduction techniques for low-latency inference.
- Collaborate with infrastructure, research, HPC, FPGA, and datacenter teams on deployment strategies that meet hardware constraints.
What they ask for
- 2+ years optimizing deep-learning inference in latency-sensitive or high-throughput production environments.
- Deep expertise in lower-level PyTorch/JAX development, strong Python/C++ skills, and understanding of mixed-precision computation.
- Proven custom GPU kernel development experience and familiarity with optimization libraries, compilers, and profiling tools.
- Deep GPU microarchitecture expertise and a data-driven record of benchmarking inference across heterogeneous architectures.
Nice to have
- Practical experience optimizing inference on specialized hardware, including FPGAs and ASICs.
Pay
$200,000-$300,000 USD annual base salary
Tools
PyTorch, JAX, Python, C++, Triton, TensorRT, ONNX, IREE
AI focus
Deep-learning inference optimization, including GPU kernel development and model quantization, pruning, and distillation.
Model-written summary · the original posting is authoritative