HS Hyosung Information Systems announced that it has successfully supported a data lake enhancement project for StradVision, a developer of AI-based vision recognition technology for autonomous vehicles.
StradVision operates an in-house AI data pipeline that connects the entire process from data collection to model training and validation, based on real-world driving data generated by vehicles operated by global customers. As its business expanded, the volume of driving data grew to several petabytes annually, increasing the need for higher performance, stability, and scalability in its AI training and validation infrastructure.
To address these requirements, HS Hyosung Information Systems carried out an AI infrastructure upgrade and data lake expansion project designed to support large-scale unstructured data processing and GPU-based analytics. The new environment allows StradVision to handle workloads more efficiently while improving cost control and scalability by extending data processing beyond a cloud-only architecture.
The project also focused on resolving storage bottlenecks in GPU-intensive AI training and validation environments. By enabling faster data access and stable file sharing for multiple engineers working simultaneously, the upgraded infrastructure improved overall system performance and increased GPU utilization, contributing to higher productivity in AI model development and testing.
Kim In-soo, Head of the Data Innovation Center at StradVision, said the data lake enhancement has helped accelerate AI model development and strengthen the company’s ability to respond to global customer demands.
Yang Jeong-gyu, CEO of HS Hyosung Information Systems, said the company will continue to support AI-driven businesses by working closely with customers to design infrastructure strategies that enable stable and efficient use of data over the long term.
