Why Join Us?
When you join AI Team at BOS Semiconductors, you will receive training and hands-on experience to master the following knowledge and skills, which will significantly benefit your career development.
In-depth Deep Learning Architecture: Study and gain a deep understanding of AI model architectures like CNNs, RNN, LSTM, and Transformers.Programming Mastery: Learn and master your skills in Python, C/C++, and other essential languages.Cutting-edge Technology: Learn about the latest in NPU architecture and how to optimize deep learning models for maximum performance.Specialized Career Path: Develop into a software engineering expert, specializing in optimizing AI models on NPUs, positioning yourself at the forefront of AI and semiconductor technology.
Company Overview:
BOS Semiconductors is a fabless startup based in South Korea, established in 2022. The company specializes in designing and developing system-on-chip (SoC) solutions. Their innovations include high-performance, low-power semiconductors, CPUs, graphics cards, and high-speed signal interfaces, all of which are essential for automotive semiconductors. Hyundai Motor Group has invested in BOS Semiconductors, anticipating collaboration towards the semiconductor technologies necessary for future mobility.BHRC (BOS Semiconductors Vietnam), established in Ho Chi Minh City in November 2022, is a key RD center comprising 40% of BOS's total RD workforce, with a primary focus on hardware and software development. The main areas of software include developing embedded software for ADAS chips, AI accelerator chips, and MCUs responsible for functional safety in autonomous vehicles. One significant opportunity for the BHRC software team is to learn about the world's most advanced NPU architecture and design the BUDA and metal-BUDA compilers for this NPU engine.The NPU vendor collaborating with BHRC is Tenstorrent, led by the renowned Jim Keller. He is famous for creating high-performance CPU architectures at Apple, Intel, AMD, and Tesla and now leads Tenstorrent, a startup developing RISC-V-based NPU engines and Transformer-based server chips. BOS's executives, former SoC developers at Samsung, have a special relationship with Jim Keller, having collaborated on several projects, including the ADAS chip for the Tesla Model 3. This relationship has allowed BOS to incorporate the world's most advanced A/I engine as a core IP in its first production chip and to take on the mission of developing the compiler for the most advance NPU architecture.
Key Responsibilities:
Develop and optimize compiler tools and runtime systems specialized for NPUs, focusing on ResNet50 and other deep learning models.Collaborate with the AI research and development team to understand the architecture of the NPU and develop strategies for optimizing AI models for this hardware.Design, implement, and optimize compiler algorithms and runtime libraries to enable efficient execution of AI models on NPUs.Work closely with the AI engineering team to integrate AI models into our automotive systems and ensure optimal performance on NPU hardware.Analyze and evaluate the performance of AI models on NPUs, identify bottlenecks, and propose solutions for optimization.Work closely with infrastructure, hardware, and do research to design solutions across the ML stack.Write clean, testable, and well-documented code to ensure long-term maintainability.Stay up to date with the latest advancements in compiler technology and AI hardware architecture to continuously improve the performance and efficiency of our AI models.
No requirement for relevant working experience
No management responsibility