Senior AI Runtime Engineer_Taipei

## About the Team

MediaTek IoT BU's Linux Platform Team develops Linux system software and distributions for the MediaTek Genio platform, working with partners, open-source communities, and upstream maintainers to deliver maintainable Yocto, Ubuntu, and Debian solutions.
We are looking for a Senior AI Engineer to lead the development and optimization of MediaTek ONNX NeronEP, our ONNX Runtime Execution Provider. This role focuses on ONNX Runtime integration, NPU acceleration, customer model performance analysis, and Debian/Ubuntu ecosystem integration.
You will help enable real-world Edge AI products across robotics, drones, and future GenAI-on-edge use cases, while contributing to ONNX Runtime upstream and broader Linux distribution integration.

## What You'll Be Doing

- Lead the development, maintenance, and performance optimization of MediaTek ONNX NeronEP, with NPU as the primary acceleration backend.
- Work closely with internal runtime, driver, compiler, and platform teams, as well as external partners, to upstream NeronEP into the official ONNX Runtime project and maintain compatibility with new ONNX Runtime releases.
- Analyze customer ONNX models, identify performance bottlenecks and functional gaps, determine whether issues come from the runtime, driver, kernel, hardware accelerator, or the model design itself, and provide concrete optimization recommendations.
- Develop model optimization, performance analysis, debug tools, and reference applications to lower the adoption barrier for MediaTek AI solutions.
- Drive the delivery of related AI toolchains as Debian/Ubuntu packages to support integration, deployment, and long-term maintenance in standard Linux distributions.
- Support customer AI projects from PoC to productization, with a focus on debugging, performance analysis, model optimization, and deployment recommendations.

## Position Info

- Location: Taipei, on-site
- Role type: Individual Contributor, focused on technical depth and cross-functional technical execution. This role does not include line management responsibilities, but requires close coordination with internal teams and external partners to drive development, integration, and product enablement.
- Organization: MediaTek IoT BU — Linux Platform Team

## What We Need to See

- 8+ years of software development experience, including 3+ years focused on AI frameworks, AI runtimes, model deployment, or related areas.
- Strong understanding of ONNX Runtime internals, with solid low-level C++ development skills and the ability to read, modify, and contribute to ONNX Runtime source code.
- Understanding of the ONNX Runtime Execution Provider architecture, including EP interfaces, Graph IR, Session, Kernel execution flow, and related core mechanisms.
- Hands-on experience with model optimization, such as quantization, graph optimization, operator fusion, memory layout tuning, calibration, or performance profiling.
- Familiarity with Python and major AI frameworks such as PyTorch, TensorFlow, and ONNX, with the ability to connect model export, optimization, and runtime deployment workflows.
- Linux system development experience, with an understanding of how Linux distributions such as Yocto, Debian, or Ubuntu are integrated.
- Strong debugging and problem decomposition skills, with the ability to analyze issues across AI frameworks, runtimes, drivers, kernels, and hardware accelerators.
- Ability to effectively leverage state-of-the-art AI tools to accelerate development, testing, analysis, and technical evaluation.
- Strong English communication skills, with the ability to collaborate with overseas third-party partners, discuss technical issues, write clear documentation, and drive alignment across teams and time zones.
- Strong cross-functional communication skills, with the ability to work with internal engineering teams, FAE teams, external partners, and customer engineering teams.

## Ways to Stand Out

- Upstream contribution experience in major open-source projects such as Linux Kernel, ONNX Runtime, ONNX, PyTorch, LLVM, or similar projects.
- Experience implementing or deeply working with ONNX Runtime Execution Providers, such as QNN EP, CoreML EP, TensorRT EP, OpenVINO EP, or similar.
- Experience bringing Mobile AI, Edge AI, or embedded AI products from PoC to mass production.
- Familiarity with open-source community collaboration, including issue triage, code review, RFCs, releases, or maintainer workflows.
- Experience with CUDA, GPU kernels, NPU kernels, or other hardware accelerator development.
- Understanding of compiler, MLIR, TVM, or AI compiler fundamentals, with the ability to discuss technical details with compiler teams.
- Experience with Debian/Ubuntu packaging or Linux distribution package maintenance.
Education: Master's Degree