Fabless automotive SOC design for autonomous driving and mobility
BOS Semiconductors designs system-on-chip solutions for autonomous vehicles, targeting autonomous driving SOCs, high-performance compute, and automotive gateway functions. The tech stack—PyTorch, TensorFlow, MLIR, GPU/TPU optimization, plus automotive safety standards (ISO 26262)—reveals a company building AI-accelerated silicon paired with software. Active projects center on NPU runtime optimization and multi-OS porting, while stated pain points (latency, power, memory) align with the core challenge of edge AI inference in vehicles. Hiring remains constrained to engineering roles in South Korea, suggesting early-stage focus on silicon design rather than scaled sales.
BOS Semiconductors is a fabless design house founded in 2022, headquartered in Seongnam (Pangyo), South Korea, with an engineering R&D expansion underway in Vietnam. The company targets automotive OEMs with system semiconductors focused on autonomous driving, HPC, and automotive gateway applications, paired with embedded software and AUTOSAR stacks. The founding team draws from senior roles at established semiconductor and automotive companies. Current operations span silicon architecture, software runtime development, and kernel optimization for next-generation mobility platforms.
PyTorch, TensorFlow, MLIR for AI workflows; GPU and TPU for acceleration; Linux, Android, QNX for OS support; C/C++ for embedded development; ISO 26262 for automotive functional safety.
Autonomous driving AI algorithm development, NPU runtime engine optimization, custom operator development, multi-OS runtime porting, multi-NPU device virtualization, and kernel/ops development for automotive SOCs.
Other companies in the same industry, closest in size