Scale-Up Ethernet Engineer
FuriosaAI is a South Korean AI semiconductor company building energy-efficient inference accelerators, servers, and software for enterprise and cloud AI deployments.
Funding history
Investors
About FuriosaAI
FuriosaAI develops the RNGD AI inference accelerator and NXT RNGD Server, alongside a software toolchain for compiling, optimizing, and deploying LLM and agentic-AI workloads.
Skills
Candidate Availability
About the Role
You will configure Ethernet endpoints, NICs and switches, and validate scale-up link-layer behavior under real AI workloads. You will analyze root causes across hardware, firmware, driver, software and fabric, validate link and optical module interoperability, and build a rack management system with observability, failure detection and test automation. You will also lead a PoC moving rack validation to OCP-standard configurations, orchestrate AI agents for lab work, and bring standards from OCP, UEC and SONiC into designs.
Requirements
- BS in Computer Science, Electrical Engineering, or related field, or equivalent practical experience
- Understanding of data-center Ethernet networks (switches, lossless/RDMA) and experience configuring and debugging network devices on Linux
- Experience developing or debugging semiconductor system software at register and memory-map level
- Hands-on experience with bring-up, integration testing, or performance analysis of rack- or cluster-scale systems
- Ability to automate test execution, device control and log collection with Python or shell scripts
- Active use of AI tools in day-to-day work and ability to verify their output
- Ability to form hypotheses from evidence and narrow down root causes through reproducible experiments
- Ability to collaborate across disciplines and document test methods and results clearly
Responsibilities
- Validate the scale-up Ethernet endpoint integrated in the accelerator SoC from pre-silicon through silicon
- Tune NICs and switch fabrics, including buffers, queues, PFC, LLR and CBFC
- Perform cross-layer root-cause analysis using packet captures, counters, logs and telemetry
- Evaluate with real AI workloads such as all-reduce and all-to-all collectives and LLM inference and training traffic
- Validate link stability and interoperability across optical modules, cables and switch ports
- Design and build a rack manager using Redfish, PLDM, gNMI/YANG and OCP specifications such as OpenRMC
- Build rack-level observability and validate reliability with fault injection
- Build and validate scale-up Ethernet racks and lead the PoC to OCP-standard configurations such as Open Rack
- Automate validation infrastructure, test execution, result collection and dashboards
- Orchestrate AI agents for bring-up, builds, testing, log analysis and documentation
- Bring OCP, UEC and SONiC standards into designs and work with vendors