AI Runtime Engineer (C/C++, Linux) + Equity

Kalray; France; Sur place; EUR 70 000 - 110 000; Plein temps

Avantages offerts par ce poste

RSU / stock options

Paid leave RTT

Mobility incentives

Generous paternity leave

Team activities

Résumé du poste

Kalray is seeking an AI Runtime / Low-Level Software Engineer to own the offload runtime for inference workloads on our custom RISC-V AI accelerator, interfacing with an x86 host. You will collaborate with AI, hardware, and compiler teams to drive the stack from emulation to production silicon.

You will develop and optimize the device lifecycle, scheduling, memory movement, and host-accelerator communication while maintaining robust CI tests and emulated environments.

Qualifications

• MSc or BAC+5 in Computer Science, Embedded Systems or related field.
• 5+ years of experience in low-level software development, embedded systems, runtime development or systems programming.
• Strong ownership mindset and autonomy to drive complex software end-to-end.

Responsabilités

• Develop and optimize the runtime for offloaded inference workloads and host-to-device communication.
• Own the performance-critical path including scheduling, memory movement and data flow.
• Profile, analyze and optimize runtime performance to reduce latency and bottlenecks.
• Design and maintain runtime APIs for AI compiler and serving layers.
• Ensure PCIe communication path works in emulation and production silicon; support CI/CD and automated tests.
• Collaborate with AI, hardware and compiler teams to improve real-world AI performance.
• Contribute to reliability and continuous improvement of the AI acceleration software stack.

Connaissances

C/C++ systems programming

Linux systems programming

Parallel programming

Embedded software

RISC-V cross-compilation

PCIe fundamentals

OpenAMP remoteproc rpmsg virtio

Linux internals

Performance profiling

Formation

MSc or BAC+5 in Computer Science/Engineering

Outils

QEMU

Description du poste

Kalray is seeking an AI Runtime / Low-Level Software Engineer to own the offload runtime for inference workloads on our custom RISC-V AI accelerator, interfacing with an x86 host. You will collaborate with AI, hardware, and compiler teams to drive the stack from emulation to production silicon.

You will develop and optimize the device lifecycle, scheduling, memory movement, and host-accelerator communication while maintaining robust CI tests and emulated environments.