Software Engineer, Runtime

FuriosaAI
Seoul
Workplace: OnsiteFull timeFunction: Software EngineeringExperience: 3+ yearsEducation: bachelorsSkills: ["Communication"]

Design and implement FuriosaAI’s low-level runtime stack for NPU hardware, pushing performance limits through DMA-based I/O, kernel execution scheduling, and asynchronous execution pipelines. Build multi-node inference primitives, including RDMA-based low-latency transfers, and develop embedded firmware (PERT) for on-device scheduling, synchronization, and resource control. Profile and tune system performance across firmware to user-space to reduce end-to-end inference latency.

Loading

Loading job details...

Preparing the role view and application actions.

FursaFursa
FuriosaAI
FuriosaAI
1 day ago

Software Engineer, Runtime

✓ Verified Job

Canonical indexed version, validated from employer's careers page.

Source: Company careers pageValidated by: Fursa AI
Last checked: 19 hours agoStatus: Live

Job Summary

Design and implement FuriosaAI’s low-level runtime stack for NPU hardware, pushing performance limits through DMA-based I/O, kernel execution scheduling, and asynchronous execution pipelines. Build multi-node inference primitives, including RDMA-based low-latency transfers, and develop embedded firmware (PERT) for on-device scheduling, synchronization, and resource control. Profile and tune system performance across firmware to user-space to reduce end-to-end inference latency.
Location: Seoul
Workplace: Onsite
Employment Type: Full time
Job Function: Software Engineering
Seniority: Mid level

Key Responsibilities

  • •Develop the low-level runtime for DMA-based I/O and kernel execution scheduling to maximize throughput and minimize end-to-end latency.
  • •Build and optimize asynchronous execution pipelines that orchestrate data movement and compute across NPU hardware.
  • •Enable multi-node inference by implementing foundational communication primitives, including RDMA-based data transfer.
  • •Develop embedded firmware (PERT) on the NPU integrated ARM core for scheduling, synchronization, and hardware resource control.
  • •Profile and tune system-level performance across firmware to user-space to remove bottlenecks in real inference workloads.

Key Requirements

  • •Bachelor’s degree in Computer Science (or equivalent) with strong systems programming experience.
  • •3+ years of systems programming in Rust, C, or C++.
  • •Strong understanding of computer architecture fundamentals including memory hierarchy, cache coherency, OS, DMA, interrupts, and MMIO.
  • •Solid communication skills for cross-team requirement gathering and technical alignment.
Experience:3+ yearsNPU hardwareAccelerator hardwareEmbedded systemsMulti-node inference
Education:Bachelor's in Computer Science
Skills:Communication
Tech Stack:RustCC++DMARDMACUDACUDA GraphsLinux kernel modulesEBPFPerfFtraceARMMMIOScatter-gather I/OZero-copyEmbedded firmwareAsynchronous executionKernel execution scheduling

Company Brief

FuriosaAI
Designs and develops data‑center AI inference accelerators (RNGD) and a full stack hardware‑software platform to deliver energy‑efficient, high‑performance AI compute for enterprise and cloud customers.
Industry: Hardware Devices
Company Size: Medium (51 to 250 employees)
Growth: Scaleup
Valuation: USD 500M to 1B
Funding: Series C
Headquarters: Seoul, South Korea
Founded: 2017
Glassdoor
Glassdoor: 4.6
WebsiteLinkedInGlassdoor