GPU Performance Engineer
Responsibilities
- Design and implement highly optimized CUDA kernels for performance-sensitive workloads.
- Analyse computational pipelines to identify bottlenecks across compute, memory, and data movement.
- Develop hardware-aware execution strategies that maximize GPU utilization and efficiency.
- Optimize inference performance through kernel tuning, memory-layout design, precision management, and workload scheduling.
- Benchmark and profile GPU applications to validate performance improvements and identify new optimization opportunities.
- Partner with researchers and engineering teams to convert computational models into efficient production implementations.
- Evaluate new GPU technologies, architectural features, and optimization techniques to enhance system performance.
- Contribute to engineering standards and best practices around performance measurement, profiling, and optimization.
Requirements
- Strong experience developing and optimizing CUDA applications in production environments.
- Solid software engineering background using C++ and performance-oriented programming techniques.
- Deep understanding of modern GPU architecture, including memory hierarchy, parallel execution models, occupancy, and performance trade-offs.
- Hands-on experience with profiling, benchmarking, and root-cause analysis of performance bottlenecks.
- Proven ability to improve application latency, throughput, and hardware efficiency.
- Strong analytical and problem-solving skills with a focus on low-level systems optimization.
- Exposure to TensorRT, Triton, TVM, ONNX Runtime, or similar acceleration frameworks.
- Familiarity with PTX-level optimization and architecture-specific performance tuning.
- Experience optimizing neural network inference workloads.
- Knowledge of kernel fusion, graph optimization, custom operators, or model compilation techniques.
- Background in scientific computing, high-performance computing (HPC), distributed systems, or other computationally intensive environments.
- Experience working with structured machine learning models and production inference platforms.
FAQs
Congratulations, we understand that taking the time to apply is a big step. When you apply, your details go directly to the consultant who is sourcing talent. Due to demand, we may not get back to all applicants that have applied. However, we always keep your resume and details on file so when we see similar roles or see skillsets that drive growth in organizations, we will always reach out to discuss opportunities.
Yes. Even if this role isn’t a perfect match, applying allows us to understand your expertise and ambitions, ensuring you're on our radar for the right opportunity when it arises.
We also work in several ways, firstly we advertise our roles available on our site, however, often due to confidentiality we may not post all. We also work with clients who are more focused on skills and understanding what is required to future-proof their business.
That's why we recommend registering your resume so you can be considered for roles that have yet to be created.
Yes, we help with resume and interview preparation. From customized support on how to optimize your resume to interview preparation and compensation negotiations, we advocate for you throughout your next career move.