Назад
Company hidden
4 дня назад

Scale-Up Ethernet Engineer (AI Infrastructure)

Формат работы
onsite
Тип работы
fulltime
Грейд
senior
Английский
b2
Страна
Taiwan
Вакансия из списка Hirify.GlobalВакансия из Hirify Global, списка международных tech-компаний
Для мэтча и отклика нужен Plus

Мэтч & Сопровод

Для мэтча с этой вакансией нужен Plus

Описание вакансии

Текст:
/
TL;DR
Scale-Up Ethernet Engineer (AI Infrastructure): Building and validating complete AI accelerator racks with Ethernet-based scale-up networking, rack management, observability, and test automation with an accent on cross-layer integration across hardware, firmware, software, switches, power, and AI workloads. Focus on configuring Ethernet endpoints and switch fabrics, diagnosing performance and stability issues under collective communication and LLM workloads, and designing reliable rack bring-up and failure-recovery systems.

Location: Hsinchu City, Taiwan

Company

hirify.global builds high-performance, energy-efficient AI compute platforms for the inference era, spanning silicon, software, compilers, and serving frameworks.

What you will do

  • Validate scale-up Ethernet endpoints integrated into accelerator SoCs from pre-silicon prototypes through silicon, including AXI, DMA, memory maps, interrupts, firmware, RTL, and diagnostic software.
  • Configure NICs and switches, tune lossless scale-up Ethernet behavior, and evaluate buffers, queues, PFC, LLR, CBFC, and purpose-built scale-up switches.
  • Reproduce transfer errors and performance bottlenecks using packet captures, counters, logs, telemetry, and real AI workloads such as all-reduce, all-to-all, LLM inference, and training.
  • Validate links and interoperability across MAC, PCS, FEC, SerDes, optical modules, DAC/AEC/AOC cables, and switch ports, including CMIS-based module checks and repeated reset and temperature testing.
  • Build a rack management system covering BMCs, switches, power shelves, BBUs, PDUs, cooling, inventory, firmware, power sequencing, topology, endpoint identity, observability, failure detection, recovery, and RAS.
  • Build complete OCP-aligned racks and automate remote control, test execution, result collection, dashboards, long-duration testing, regression validation, and multi-agent lab workflows.

Requirements

  • BS in Computer Science, Electrical Engineering, or a related field, or equivalent practical experience.
  • Understanding of data-center Ethernet networks, switches, lossless networking, and RDMA, with experience configuring and debugging network devices or Linux systems.
  • Experience developing or debugging semiconductor system software at the register and memory-map level, such as SoC firmware, device drivers, silicon bring-up, or diagnostic software.
  • Hands-on experience with bring-up, integration testing, or performance analysis of rack- or cluster-scale systems combining servers, switches, and accelerators.
  • Ability to automate test execution, device control, and log collection with Python or shell scripts.
  • Active day-to-day use of AI tools, LLMs, or coding agents, with the ability to verify their output against evidence.
  • Ability to form hypotheses from packets, logs, and performance counters and narrow root causes through reproducible experiments.
  • Ability to collaborate across disciplines and document test methods and results clearly.

Nice to have

  • Experience with 100G–800G Ethernet, MAC/PCS/FEC/SerDes, QSFP-DD or OSFP modules, DAC/AEC/AOC cables, CMIS, interoperability testing, and rack-scale cabling.
  • Experience configuring lossless Ethernet fabrics and congestion control, including PFC, ECN, buffers, and queues, or working with Broadcom Tomahawk switches and switch NOS architectures.
  • Experience analyzing DMA, remote memory access, and collective communication on NPU/GPU systems, or debugging Ethernet controller IP and SoC interactions at C/C++ and RTL levels.
  • Experience with BMC/OpenBMC, Redfish, PLDM/MCTP, OpenRMC, rack or fleet management, firmware updates, power and thermal control, and failure recovery.
  • Experience building hardware test labs, CI regression systems, telemetry, long-duration stability infrastructure, or multi-agent device-control workflows.
  • Experience with OCP Open Rack, ORv3/ORW, 48 V or 800 VDC power, power shelves, BBUs, ODM rack bring-up, OCP ESUN/SUE-T, UEC, UALink, or SONiC communities.

Culture & Benefits

  • Work on full-stack AI infrastructure spanning silicon, software, compilers, and serving frameworks.
  • Contribute to RNGD, a chip in mass production with TSMC and deployed in enterprise workloads.
  • Collaborate with industry partners including TSMC, Broadcom, SK Hynix, and GUC.
  • Work in a high-trust, low-ego environment that values autonomy, intellectual curiosity, and shared ambition.
  • Use AI agents to distribute bring-up, builds, testing, log analysis, and documentation while validating their results against evidence.

Будьте осторожны

Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →