Fintegrity
Back to Global Search

d-Matrix

d-matrix.ai
Official listings

Senior Staff Inference Performance Engineer

US · Santa Clara · Santa Clara, Remote · Remote · Full-time · Senior

High payRemote worldwideSeniorSalary disclosedOpen to China
PythonC/C++LinuxNUMAKubernetesLLM InferenceSpeculative DecodingKV-CacheCUDAProfiling

USD 175,000-265,000/yr ≈ ¥1,242,500-1,881,500/yr

AI summary

About the company

d-Matrix是一家开发内存中心架构的AI基础设施公司,打造从芯片、软件到机架级系统的产品,推动AI推理的高效可及。

Role overview

该岗位隶属于d-Matrix的推理云工程团队,负责端到端的推理工作负载性能,需诊断全服务路径的性能问题,确保工作负载性能不低于裸金属水平,团队正攻克AI基础设施的核心挑战,采用内存中心架构打造高效能AI解决方案。

Responsibilities

  • 诊断全服务路径的性能问题,覆盖编排到推理内部各环节
  • 隔离问题至对应层与负责人,决定修复、转交或两者结合
  • 编写可复现的问题说明,助力跨栈团队协作
  • 保障编排工作负载的吞吐量、延迟与稳定性,缩小与裸金属的性能差距
  • 评估从单部署到 fleet 运营的扩展需求
  • 在编排层配置工作负载节点要求,确保高性能部署为默认
  • 构建轻量级遥测与基准测试工具,隔离或转交问题
  • 将重复调查转化为操作手册,指导工程师采用 profiling 方法

Requirements

  • 8年以上性能或系统工程经验,解决跨栈的复杂性能回归问题
  • 有在陌生、文档稀疏系统中通过 profiling 定位性能开销的经验
  • 具备Linux、NUMA、CPU pinning等系统底层基础
  • 有Kubernetes上性能敏感工作负载的运行经验
  • 了解LLM推理服务的核心技术,如batching、KV-Cache等
  • 精通Python和C/C++,可跨栈调试与代码修改
  • 具备清晰的技术写作能力,供跨团队使用

Nice to have

  • 有vLLM、SGLang等LLM推理框架使用经验
  • 有构建或调优推测解码、异构服务架构的经验
  • 有KV-Cache管理、请求调度等调优经验
  • 有GPU编程与profiling经验,如CUDA、Nsight等
  • 有Rust经验或创业公司工作经验

Pay & benefits

  • 薪资范围175K-265K美元
  • 提供股权
  • 提供奖金

Register to apply directly, open the original posting and save jobs

Sign up free

This is not an official page of d-Matrix. The listing is gathered automatically from the employer's official careers page or a job board's public feed (see sources on the right). Fintegrity has no agency relationship with this employer and has not verified its credentials. Chinese titles and summaries are AI-generated; the original text prevails. Apply on the source page. We never charge job seekers. Ranking & display rules

Excerpt

Join d-Matrix and help build the future of AI inference. We’ve pioneered a memory-centric architecture that brings compute directly into memory, making AI faster, more energy efficient, and accessible on a global scale. From chips and software to rack-scale systems, our team is tackling some of AI…

Sign up free

The summary and highlights are AI-generated and may contain errors; the original posting prevails.

Sources or employers who want a posting removed or corrected can email contact@fintegrity.cn; we act within 24 hours.

Senior Staff Inference Performance Engineer · d-Matrix · Global Search | Fintegrity