ML Runtime Engineer — Edge Inference & Equity

ATX Venture Partners; Occitanie; Sur place; EUR 85 000 - 110 000; Plein temps

Avantages offerts par ce poste

Equity

Paid time off

Health insurance

Lunch vouchers

Cross-office travel

Off-sites

Relocation assistance

Résumé du poste

ATX Venture Partners is seeking an experienced engineer to own model compilation for accelerator targets, turning ONNX models into optimized engines while respecting on-device power and thermal budgets.

The role focuses on building the ground-side compile and delivery path, performance benchmarking, and contributing to the Space Inference Engine and SDK. You will coordinate with the OBSW/Runtime team to ensure robust, scalable model builds.

Qualifications

• Strong C++ and Python programming skills.
• Experience with model compilation and graph compilers (TensorRT).
• Experience with ONNX Runtime and performance optimization.
• Experience with embedded/edge GPU deployment (NVIDIA Jetson).
• Experience benchmarking and profiling with perf tools.
• CI/CD experience including hardware-in-the-loop testing.

Responsabilités

• Own model compilation: turn partner ONNX models into optimized engines per accelerator.
• Build the ground-side compile & delivery path; define how a mismatched engine is detected/rejected on-node.
• Build performance tooling: benchmarking, profiling, operator-coverage matrices, and budget validation.
• Contribute to Space Inference Engine (execution providers) and the SDK (model build & packaging). Coordination of OBSW / Runtime team.

Connaissances

C++ and Python

TensorRT

ONNX Runtime & perf optimization

Embedded / edge GPU deployment

Benchmarking & profiling

CI/CD with hardware-in-the-loop

Outils

Hailo SDK

Meson build

Yocto / minimal Debian images

Description du poste

ATX Venture Partners is seeking an experienced engineer to own model compilation for accelerator targets, turning ONNX models into optimized engines while respecting on-device power and thermal budgets.

The role focuses on building the ground-side compile and delivery path, performance benchmarking, and contributing to the Space Inference Engine and SDK. You will coordinate with the OBSW/Runtime team to ensure robust, scalable model builds.