NPU Runtime SW Engineer

Job Title

NPU Runtime Software Engineer

Work Location

Seongnam-si, Korea

✔️ Key Responsibilities

• Designing a runtime module that interfaces compiler and driver modules
• Design and implement user-level APIs, add support for various language bindings, deploy ML models to SDK, and maintain user documentation
• Conduct benchmarking and profiling to evaluate the existing runtime system's performance and implement optimizations to enhance the overall system performance of NPU products

✔️ Requirements

• Bachelor's or Master's degree in computer science, electrical engineering, or a related field
• Comprehensive understanding of deep learning models and their applications in vision, natural language processing, speech recognition, and other domains
• Familiarity with system software, including compilers, runtimes, drivers, firmware, etc
• Proficient in programming languages: C++ and Python
• Knowledge of data structures, algorithms, and OOP design patterns
• Strong written and verbal communication skills

✔️ Preferred Qualifications

• Hands-on experience with AI accelerator (e.g., GPU) driver APIs and runtimes
• Exposure to ML frameworks such as PyTorch, TensorFlow, ONNXRuntime, TensorRT, and their respective optimization techniques
• Solid understanding of operating systems, resource management, and high-performance computing principles
• Deep expertise in modern C++ and its advanced features for writing efficient, high-performance code
• Experience with multithreading and parallel programming
• Proven experience in porting and accelerating deep learning models on various hardware platforms, including x86 CPUs with SSE/AVX instructions, ARM CPUs with NEON instructions, and heterogeneous SoCs with CPU, GPU, NPU, and DSP components

Contact

Jinyeong Jeong, Consultant

02-6200-9736

jinyeong.jeong@adecco.com