We are seeking a talented software engineer with expertise in ML accelerators.
This position involves designing and developing the system software at the lowest level, including device drivers and runtime, to efficiently support AI applications.

The role also includes defining and implementing OS interfaces and integration, as well as contributing to the inter-accelerator infrastructure. The ideal candidate will work across multiple components of the system software stack, focusing on system-level performance optimization and ensuring seamless hardware-software interaction.

Responsibilities:

• Develop, optimise and test low-level system software, including kernel-space device drivers and user-space runtime
• Create user-facing tools such as profilers, system monitors, and management utilities
• Design and implement interconnect infrastructure for accelerator communication
• Collaborate with the hardware team to define hardware-software interfaces for FPGA and ASIC based platforms
• Support hardware debugging and testing
• Participate in ML hardware requirements analysis, architecture iteration, and optimization to develop high-performance accelerators.

Qualifications:

• 3+ years of system software development experience
• Proficiency in developing drivers and profiling tools for custom hardware platforms, such as FPGA, GPU, or NPU
• Strong knowledge of ML accelerator system software stack, with expertise in mechanisms like asynchronous launches, asynchronous DMA, multi-stream, and multi-event operations
• Hands-on experience in Linux device driver development, particularly for ML accelerator, PCIe and network cards
• Strong coding skills in C and Shell scripting
• Excellent problem-solving skills, self-motivated, and capable of working collaboratively in a cross-functional team

Preferred Skills & Qualifications:

• Proven experience in developing complex software systems from the ground up
• Experience with ML hardware simulation, emulation and bring-up
• Expertise in system software development for interconnect communication
• Experience with CUDA-like system software
• Knowledge of AI inference and ML architectures