You will work as part of an established and growing research team, and support in the productionisation of large/complex models for maximum trading impact. You’ll closely collaborate with researchers, hardware and software engineers, and leverage your knowledge of GPU hardware and libraries to drive the optimisation of inference & hardware utilisation.
Your Core Responsibilities:
- Leverage your knowledge of GPU architecture & libraries to optimise model structure, library use and software-hardware integration
- Deliver extensions, fixes and improvements on top of base CUDNN & TensorRT to support performant evaluation
- Build C++ and CUDA based performance enhancement libraries for common deep learning libraries like Pytorch and JAX
Your Skills and Experience:
- MS degree in CS or similar fields or equivalent experience
- 3+ years of relevant work experience
- Experienced in fundamental libraries for accelerating ML workflows, like CUDNN/TensorRT, ROCm, OpenVino or OpenPPL.(understanding of one or more ML communication frameworks like NCCL, is an advantage)
- Background in deep learning fundamentals and common deep learning software, especially PyTorch
- Experienced in C++ & CUDA. Experience in Python is also highly desirable
About Us
IMC is a leading trading firm, known worldwide for our advanced, low-latency technology and world-class execution capabilities. Over the past 30 years, we’ve been a stabilizing force in the financial markets – providing the essential liquidity our counterparties depend on. Across offices in the US, Europe, and Asia Pacific, our talented employees are united by our entrepreneurial spirit, exceptional culture, and commitment to giving back. It's a strong foundation that allows us to grow and add new capabilities, year after year. From entering dynamic new markets, to developing a state-of-the-art research environment and diversifying our trading strategies, we dare to imagine what could be and work together to make it happen.