Remote AI Accelerator Framework Engineer - Principal

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

Benefits offered by this job

Premium medical health care

Generous paid time off

Office amenities and snacks

Remote work policy with reimbursement

Job summary

Ampere is a semiconductor design company forging a new era of high-performance, energy-efficient AI compute. As an AI Accelerator Software Principal Engineer – Framework Integration, you will own the strategy and delivery for DL inference across Ampere accelerators, enabling PyTorch/ONNX/llama.cpp integration that drives real-world performance from data centers to edge.

You will lead cross-team efforts, architect end-to-end acceleration across SW/HW, and mentor teams to raise the bar on quality

Qualifications

• Education & experience: BS CS/CE/EE or related field + 8y; or MS +6y; or PhD +3y.
• Deep framework expertise: PyTorch, ONNX, and llama.cpp integration with performance focus.
• Linux accelerator runtime/driver experience (preferred).
• Strong systems programming + performance tuning: Python and C/C++.

Responsibilities

• Framework integration leadership for PyTorch/ONNX/llama.cpp into Ampere backend.
• Drive end-to-end SW/HW acceleration across the stack including inference serving and runtime.
• Model enablement: optimize performance and accuracy for popular frameworks and stacks.
• HW/SW co-design to maximize efficiency and scalability across cores and memory.

Skills

Python

C/C++

PyTorch

ONNX

llama.cpp

Performance tuning

Framework integration

Education

BS in Computer Science/Engineering

MS in related field

PhD in related field

Tools

Linux

Linux drivers

Runtime libraries

Job description

Ampere is a semiconductor design company forging a new era of high-performance, energy-efficient AI compute. As an AI Accelerator Software Principal Engineer – Framework Integration, you will own the strategy and delivery for DL inference across Ampere accelerators, enabling PyTorch/ONNX/llama.cpp integration that drives real-world performance from data centers to edge.

You will lead cross-team efforts, architect end-to-end acceleration across SW/HW, and mentor teams to raise the bar on quality