Were looking for a
Principal Software Engineer
with 8+ years of experience to lead the design and implementation of our host-side PCIe drivers. You will architect and build a high-performance, multi-OS driver solution that supports complex hardware topologies. This role requires deep expertise in low-level system software, cross-platform driver development, and a proven ability to lead technical initiatives.
Key Responsibilities
-
Multi-OS PCIe Driver Development:
You will serve as the technical lead for designing and implementing high-performance PCIe kernel-mode drivers
for both Linux
and Windows
. Your work will be crucial in developing a robust architecture that supports multiple endpoints
across multiple cards
and ensures reliable communication. -
User-Space Library & PyTorch Runtime Integration:
Architect and implement a high-performance user-space library
that directly interfaces with the kernel-mode driver. This library will be the primary communication channel for the PyTorch runtime
, providing a seamless and efficient way to send commands and data to the device. -
PCIe Switch Management:
Architect and implement the driver logic to support complex hardware topologies involving multiple PCIe cards connected over an external PCIe switch
. Youll be responsible for the sophisticated driver logic required to identify and correctly set up the PCI device address
for each individual PCIe device on each card. -
Performance & Reliability:
Take ownership of the drivers performance, ensuring high-bandwidth and low-latency data transfers. Youll be responsible for the stability and reliability of the system, designing robust error handling and recovery mechanisms. -
Cross-Functional Leadership:
Collaborate closely with hardware, firmware, and user-space software teams. You will be a key voice in defining system-wide architectural requirements and ensuring a cohesive, end-to-end solution. -
Mentorship:
Guide and mentor other engineers, providing technical direction and fostering a culture of excellence in low-level software development.
Required Qualifications
Kernel-Mode Expertise:
-
PCIe Architecture:
Expert-level understanding of the PCIe specification
, including complex topologies involving switches
, bus numbering, and device address management. -
Problem-Solving:
Exceptional debugging skills and a proven track record of solving the most challenging system-level problems. -
Leadership:
Demonstrated ability to lead technical projects and mentor engineers.
Education:
-
Experience:
8+ years of hands-on experience in low-level system software development. -
Kernel-Mode Expertise:
Deep, hands-on experience designing and developing kernel-mode drivers
for both Linux
and Windows
. -
Linux:
In-depth knowledge of the Linux kernel driver model and debugging tools. -
Windows:
Extensive experience with the Windows Driver Kit ( WDK
) and the WDM/KMDF
driver models.
-
PCIe Architecture:
Expert-level understanding of the PCIe specification
, including complex topologies involving switches
, bus numbering, and device address management. -
Problem-Solving:
Exceptional debugging skills and a proven track record of solving the most challenging system-level problems. -
Leadership:
Demonstrated ability to lead technical projects and mentor engineers. -
Education:
Bachelors or Masters degree in Computer Science, Electrical Engineering, or a related field.