Senior Kernel Engineer: Custom Linux for Confidential AI

Chikara HR; Paris; Sur place; EUR 85 000 - 100 000; Plein temps

Avantages offerts par ce poste

Own kernel equity

Publishable work

Leadership track in systems

Training budget

Flexible working

Equity ( BSPCE )

Résumé du poste

Nexus in Paris is building a confidential AI platform with a custom Linux kernel at its core. You will be among the first hires to design a stripped-down kernel that bridges CPU-encrypted memory with GPU compute, enabling secure data processing.

The role is on-site in Paris with a team that works in English and clients to speak French. Seniority is required to own end-to-end hard systems work; CUDA is a bonus and BSPCE equity is offered.

Qualifications

• You write low-level C and C++ every day: manual memory management, pointers, the systems layer.
• You understand Linux internals, DMA, PCIe and how the hardware actually behaves.
• You have made systems faster and can show the profiling to prove it.
• You are senior enough to own a hard systems component end to end at an early-stage company.
• CUDA and GPU programming are a bonus, not a must. A strong systems engineer can pick them up here.
• This is a purely low-level systems role. Full-stack and web backgrounds are not the fit.

Responsabilités

• Build and maintain a stripped-down custom Linux kernel that bridges CPU-encrypted memory with GPU compute.
• Make the GPU read encrypted memory safely, and keep the I/O strictly air-gapped.
• Provision and tune NVIDIA GPUs, down to PCIe passthrough and time-slicing.
• Wring latency out of a tightly constrained system, one context switch at a time.
• Contribute to open source where the work allows.

Connaissances

Low-level C/C++

Linux internals

DMA & PCIe

Performance profiling

Systems engineering

CUDA (bonus)

Outils

CUDA

Description du poste

Nexus in Paris is building a confidential AI platform with a custom Linux kernel at its core. You will be among the first hires to design a stripped-down kernel that bridges CPU-encrypted memory with GPU compute, enabling secure data processing.

The role is on-site in Paris with a team that works in English and clients to speak French. Seniority is required to own end-to-end hard systems work; CUDA is a bonus and BSPCE equity is offered.