Principal AI Accelerator Runtime Engineer

Ampere Computing; Warszawa; On-site; PLN 302,000 - 453,000; Full time

Benefits offered by this job

Premium medical care

Generous PTO

Gym and sauna access

Flexible hours

Remote work reimbursement

Sports card

Job summary

Ampere Computing is hiring an AI Accelerator Principal Software Engineer to lead the design and optimization of AI runtime software for its deep learning accelerators. You will shape execution, memory management, and orchestration across the SW/HW stack to deliver high throughput and low latency for diverse model types and frameworks.

You will collaborate with hardware and platform teams to ensure robust deployment, performance verification, and scalable integration in production workloads.

Qualifications

• Strong background in developing user‑mode drivers and runtime libraries for GPUs or DL accelerators on Linux or RTOS.
• Hands‑on experience with AI frameworks and runtime integration (e.g., PyTorch, ONNX).
• Excellent C/C++ systems programming, memory management, threading, and performance profiling.

Responsibilities

• Design, develop, and optimize an AI Runtime Library for Ampere accelerators to support diverse models and frameworks.
• Own end‑to‑end acceleration paths across the SW/HW stack, including inference serving, graph/IR execution, and memory management.
• Drive HW/SW co‑design and optimization for throughput, latency, and memory efficiency.
• Collaborate with hardware and systems teams to ensure runtime alignment with accelerator capabilities.
• Integrate runtime components into Ampere platform stacks for production‑like workloads.

Skills

C/C++

Systems programming

Performance profiling

AI framework enablement

PyTorch

llama.cpp

ONNX

GPU runtime libraries

Education

BS in CS/CE/EE or software eng

MS in CS/CE/EE or software eng

PhD in relevant field

Tools

Linux

RTOS

Job description

Ampere Computing is hiring an AI Accelerator Principal Software Engineer to lead the design and optimization of AI runtime software for its deep learning accelerators. You will shape execution, memory management, and orchestration across the SW/HW stack to deliver high throughput and low latency for diverse model types and frameworks.

You will collaborate with hardware and platform teams to ensure robust deployment, performance verification, and scalable integration in production workloads.