Two Sigma
GPU Performance Engineer
$165K–$300K
NY New York · Full-time · Posted Oct 28, 2026
- Experience
- Minimum 1 year required; 4-10 years preferred.
- Education
- BS or MS in Science, Technology, Engineering or Math.
About the job
Designs and optimizes GPU computation for financial workloads as Two Sigma transitions from CPU-centric to GPU-accelerated computing.
What you'll do
- Design and optimize GPU kernels and workloads for throughput, latency, and memory efficiency.
- Develop precision-management procedures for FP8/FP4 financial applications and profile workloads with NVIDIA tools.
- Build reusable GPU libraries and evaluate GPU-accelerated libraries for financial use cases.
What they ask for
- BS or MS in Science, Technology, Engineering or Math.
- Expert-level CUDA programming and deep understanding of GPU architecture.
- Experience profiling and optimizing GPU workloads; strong C++ and Python skills.
- Familiarity with mixed-precision computation and numerical stability, and a record of delivering speedups on real workloads.
Nice to have
- Background in HPC, scientific computing, or computational finance.
- Experience with multi-GPU and multi-node programming (NCCL, MPI), or familiarity with RAPIDS and cuDF.
Pay
$165,000 and $300,000 USD base salary
Tools
CUDA, C++, Python, NVIDIA tooling (Nsight Systems, Nsight Compute), RAPIDS, CUTLASS, cuBLAS, TensorRT
AI focus
Supports the firm's AI Innovation and GPU acceleration strategy, including FP8/FP4 training and inference for financial applications.
Model-written summary · the original posting is authoritative