How Woven by Toyota Saved Millions with Scaled Autonomous Driving from Union.ai
The Challenge
Orchestrating AI at Autonomous Scale
Woven by Toyota (WbyT), a part of the Toyota Group, is challenging the current state of mobility through the automated driving (AD/ADAS) technology, and the automotive software development platform Arene OS. These AI workloads are among the most complex in the world, requiring…
- Real-time decision-making
- Massive, petabyte-scale data processing
- GPU-intensive training pipelines
- Strict safety and regulatory compliance
WbyT initially relied on Flyte for workflow orchestration. But as their ambitions grew, maintaining internal infrastructure became a bottleneck. WbyT realized that they needed more than just orchestration—they needed a reliable, enterprise-grade solution to power compound AI systems at scale. That meant high-throughput data handling, GPU compute management, reproducibility for model training, and seamless integration with internal tools.
The Solution
Union.ai – A Unified AI Developer Suite for Compound AI Systems
In 2023, WbyT migrated to Union.ai, a fully managed, production-grade platform built to support the most advanced AI workloads in the industry. Union.ai delivered far more than managed Flyte - it became WbyT’s foundational infrastructure for developing, orchestrating, and scaling compound AI systems.
“Union.ai support has proven to be critical for us at the time that we needed to significantly scale up our data processing needs to meet major milestones. Union.ai’s wealth of expertise in this domain has enabled us to focus our efforts on key ADAS-related functionalities, move fast, and rely on Union.ai to deliver data at scale.” —Alborz Alavian, Senior Engineering Manager, Woven by Toyota
Today, Union.ai powers WbyT’s data annotation, perception labeling, dynamic scene processing, and AI model training pipelines. Union.ai manages massive workloads that have processed petabytes of data and hundreds of thousands node hours of computing for WbyT’s autonomous vehicle programs. Crucially, it abstracts infrastructure complexity so that developers and ML engineers can focus on building smarter, safer systems—not managing pipelines and maintaining infrastructure.
The Impact
Accelerated Innovation, Lower AI Costs, and Massive Scale
Union.ai’s business impact on WbyT’s ADAS and AI initiatives has been transformative:
- Over 20x faster ML iteration cycles – The speed of model iteration and deployment was dramatically improved.
- Millions of dollars in annual cost savings – Through advanced scheduling, spot instance optimization, and compute caching, Union.ai dramatically reduced cloud infrastructure spend.
- 1000s of parallel workers – Massive parallelism enabled the team to run large-scale workloads and unlock real-time development velocity.
- Massively scalable backfilling and reproducibility – Union.ai enables seamless recomputation of historical sensor data, supporting continuous improvement of AV models.
- Reduced DevOps maintenance overhead – With Union.ai, WbyT offloaded Flyte maintenance entirely, allowing engineers to focus on innovation instead of infrastructure.
Why Compound AI Systems Matter for Toyota's Future
As WbyT continues to invest in the future of mobility, Union.ai provides its foundation for compound AI systems—where data processing, model training, real-time inference, and infrastructure scale come together seamlessly.
Ready to accelerate your AI like Woven by Toyota?
Union.ai is the AI developer suite for compound AI systems. If your team is pushing the boundaries of autonomous systems, let’s talk.
<div class="button-group is-center"><a class="button" href="https://meetings.hubspot.com/hello-union/discover-union">Book a meeting with Union.ai</a></div>