High-Performance ML Backend Engineer

Bytedance; San Jose, Northern (CA, KY); Hybrid; USD 180,000 - 240,000; Full time

Job summary

ByteDance, a global tech company behind TikTok, is seeking talent for its Data AML mid-platform team in San Jose. The role focuses on optimizing the computational performance of recommendation/inference systems, enabling powerful ML compute for internal businesses across multiple products including Douyin and Xigua.

Successful candidates will design high-performance kernels, drive model compilation optimization, and collaborate with hardware/software teams to diagnose bottlenecks and implement

Qualifications

• Bachelor's/Master's degree in Software Development, Computer Science, Computer Engineering, or related technical discipline.
• Familiar with mainstream model compilation stacks (TVM, MLIR, XLA, etc.) with development and optimization experience.
• Proficient in C/C++ development; familiar with CPU/GPU architecture and cache mechanisms; practical high-performance kernel development experience.
• Understanding of deep learning frameworks (TensorFlow, PyTorch, OneFlow, etc.) and their computational graphs, inference/training execution, and graph optimization/compilation.

Responsibilities

• Optimize computational performance of ByteDance's recommendation mid-platform models; tune for inference/training bottlenecks in business scenarios; improve kernel computing utilization.
• Lead design/development of high-performance kernel libraries, covering general-purpose and business-specific kernels, including parallelism for computation and communication.
• Advance model compilation optimization via graph optimization, kernel fusion, scheduling, and code generation; build/improve mid-platform model compilation system.
• Collaborate with business/algorithm teams to identify performance issues; provide full-stack performance analysis, bottleneck diagnosis, and optimization solutions; consolidate general-purpose performance components, toolchains, and platform capabilities.

Job description

ByteDance, a global tech company behind TikTok, is seeking talent for its Data AML mid-platform team in San Jose. The role focuses on optimizing the computational performance of recommendation/inference systems, enabling powerful ML compute for internal businesses across multiple products including Douyin and Xigua.

Successful candidates will design high-performance kernels, drive model compilation optimization, and collaborate with hardware/software teams to diagnose bottlenecks and implement