STRADVISION: Powering AI-Driven Automotive Innovation with Scalable Storage
How Qumulo enables STRADVISION to manage explosive data growth and accelerate autonomous driving development
STRADVISION is redefining the future of autonomous vehicles with advanced AI-powered vision recognition software. Headquartered in Seoul, with data operations in Pohang, the company develops SVNet—its flagship object recognition technology used in autonomous vehicles and Advanced Driver-Assistance Systems (ADAS).
With a mission to accelerate the development of safer, smarter transportation, STRADVISION generates and processes enormous volumes of unstructured data—over 2 petabytes and growing. But as data volumes surged, so did the limitations of their legacy storage systems.
“Our NAS environment was fragmented, inefficient, and unreliable,” said Euncheol Yeo, Head of IT Infrastructure at STRADVISION. “Managing over 20 distributed Synology NAS units caused performance bottlenecks and frequent system failures, especially under the load of hundreds of simultaneous users.”
System freezes, manual cold reboots, and constrained network performance slowed production and jeopardized efficiency. The team recognized the need for a solution that could handle rapid growth, support large-scale concurrent access, and recover from failures quickly.
They turned to Qumulo.
Qumulo delivered immediate results. We observed stable performance, simplified operations, and the capability to monitor and analyze data in real-time. It was a game-changer.
STRADVISION deployed two on-premises Qumulo clusters and integrated Qumulo’s cloud-native CNQ platform with AWS to establish a hybrid cloud file storage environment. The transition from Synology to Qumulo took less than a month and was supported by strong collaboration between Qumulo, local partners, and STRADVISION’s technical teams.
“Qumulo delivered immediate results. We observed stable performance, simplified operations, and the capability to monitor and analyze data in real-time. It was a game-changer.”
As user demand continued to grow, so did the deployment. STRADVISION expanded from 13 to 18 Qumulo nodes, now running across two clusters with over 2PB of data.
Today, Qumulo supports mission-critical workflows for hundreds of concurrent users. The development of safe AI vision software for single- and multi-sensor autonomous driving is a data-intensive process, requiring storage and data management systems that can efficiently support large-scale CPU and GPU cluster workloads. These include data ingest, preprocessing, AI model training, testing, and full system validation. Qumulo delivers the scalability and performance STRADVISION relies on to drive continuous innovation in the future of autonomous mobility.
Even with hundreds of users accessing data at the same time, Qumulo maintains stability and speed. It’s helped us significantly reduce both work time and operational costs.
Business Impact
- Accelerated Model Development over prior solution across 100s of developers across PBs in the cloud on and on-premises.
- Enhanced Operations Efficiencies with centralized data management including real time analytics and monitoring.
- Enabled faster issue resolution and minimized downtime through exceptional 24/7 access to Qumulo engineers.
Looking Ahead
With Qumulo, STRADVISION is building the data foundation for a safer, autonomous future—one object, one frame, one innovation at a time.
“If you’re working with massive volumes of unstructured data, Qumulo is the best solution. Our only regret is not switching sooner. The 24/7 private Slack channel and proactive Zoom support from both Pre-Sales and CSE engineers make Qumulo unlike anything we’ve experienced. They’ve stood by us during everything from data center outages to hardware failures—resolving issues quickly and thoroughly.”
Industry:
Automotive, Artificial Intelligence
Use Case:
Unstructured data management at scale for autonomous vehicle development
Deployment:
On-prem + CNQ with AWS hybrid deployment
Location:
Seoul, South Korea
Company Overview:
STRADVISION, a leader in AI-based vision recognition software for autonomous driving, needed a reliable, scalable data platform to handle its explosive data growth and support mission-critical workloads. With over 2PB of unstructured data and hundreds of users working simultaneously, legacy NAS systems couldn’t keep up.
Why Qumulo:
- Replaced 20+ fragmented NAS units with centralized, high-performance storage
- Real-time visibility into data operations and system health
- Seamless hybrid cloud deployment with AWS
- Expanded from 13 to 18 nodes in under a month
- Exceptional 24/7 support, including dedicated Slack and Zoom access
Results:
- Enabled reliable performance for AI training and high-resolution image storage
- Reduced operational overhead and unplanned downtime
- Empowered technical teams with real-time analytics and monitoring
- Improved efficiency, scalability, and cost-effectiveness