Lead GPU System Software Architect for Data-Center Compute

ByteDance; San Jose (CA); On-site; USD 230,000 - 280,000; Full time

Job summary

ByteDance in San Jose is seeking a hands-on leader to drive the low-level GPU software stack for data-center compute. You will define architecture, set technical standards, and guide cross-functional teams across silicon, firmware, kernel, runtimes, and fleet management.

You will remain deeply involved through prototyping, critical-path reviews, performance analysis, and high-severity investigations, communicating decisions to engineers and executives alike.

Qualifications

• Bachelor's degree in CS/CE/EE or equivalent practical experience.
• 5+ years hands-on GPU/accelerator/kernel/system software with leadership on complex hardware/software programs.
• Proven ability to lead hardware/software programs from concept to production.
• Deep understanding of accelerator architecture, memory, interconnects, OS, device management, and reliability.
• Experience coordinating dependencies across multiple engineering teams.
• Strong coding, design review, performance analysis, and debugging with hands-on contributions.

Responsibilities

• Define architecture and multi-year technical roadmap for GPU system software across driver, runtime, firmware-interface, management, observability, and reliability layers.
• Set priorities and standards for accelerator enablement balancing product commitments with compatibility, performance, serviceability, and long-term maintenance.
• Own end-to-end integration from architecture and pre-silicon planning through bring-up, qualification, GA, fleet monitoring, and sustained operation.
• Lead cross-functional execution among silicon, firmware, kernel, compiler, library, ML framework, server, network, storage, validation, and production teams.
• Resolve ambiguous system-level tradeoffs involving APIs, resource management, memory and interconnect topology, telemetry, recovery, security, and workload performance.
• Remain technically involved through prototypes, critical-path code and design reviews, performance analysis, and leadership during high-severity failure investigations.
• Define measurable release and reliability criteria, including qualification coverage, regression thresholds, fault containment, automated repair, and fleet-health indicators.
• Mentor engineers, raise the quality of architecture and debugging practices, and communicate technical decisions to both specialist and executive audiences.

Skills

GPU software leadership

Hands-on system debugging

Performance analysis

Cross-team coordination

Architecture guidance

Education

Bachelor's degree in CS/CE/EE

Tools

Kernel drivers

Observability tooling

Job description

ByteDance in San Jose is seeking a hands-on leader to drive the low-level GPU software stack for data-center compute. You will define architecture, set technical standards, and guide cross-functional teams across silicon, firmware, kernel, runtimes, and fleet management.

You will remain deeply involved through prototyping, critical-path reviews, performance analysis, and high-severity investigations, communicating decisions to engineers and executives alike.