dcoded

CUDA Backend Developer - Contract Remote | £400-£500 per day | Outside IR35 | 6-Month Initial Engagement Overview We're looking for an experienced CUDA Backend Developer to join a high-performance engineering team working on GPU-accelerated simulation and AI workloads. You'll collaborate with C++ systems engineers and research scientists to design, implement, and optimize GPU-intensive Back End modules that push the limits of performance and scalability. What You'll Do Design and implement GPU kernels in CUDA C, focusing on: Kernel fusion On-device operations GPU memory optimization Build and use profiling tools (eg, Nsight) to measure and improve GPU utilization, inference latency, and training throughput. Optimize custom models for deployment with TensorRT or similar inference engines. Integrate GPU functionality into Back End APIs and orchestration layers. Work closely with research and engineering teams to translate models into performant CUDA implementations. What We're Looking For Strong experience in C++ (11/14/17) and CUDA C programming. Proven track record using GPUs for compute-intensive applications in production environments. Hands-on with CUDA profiling, debugging, and Kernel optimization. Deep understanding of multi-threaded/multi-process architectures and Linux systems development. Proficiency in low-level I/O, memory management, and performance tuning Nice to Have Experience with distributed training/inference pipelines. Familiarity with Docker and Kubernetes. Exposure to Embedded systems or hardware-level software integration. Start: ASAP Duration: 6 months (strong potential to extend) Location: Remote (UK or EU-based preferred) IR35: Outside If you're a GPU performance enthusiast who thrives on complex Back End challenges and wants to contribute to cutting-edge AI systems, we'd love to hear from you. Apply now or get in touch directly for a confidential conversation.