Background
A leading Tier 1 automotive supplier with over 162,000 employees worldwide supplying critical components to major automakers. Denso’s Hardware and Embedded Software Development team designs Electric Power Steering (EPS) software and tests high-performance embedded systems for vehicle steering, ensuring safety, reliability, and efficiency. However, this team has traditionally operated in an on-premises environment, with no prior adoption of cloud-based services. The team struggled with lengthy testing cycles for their EPS software, causing recalls, delays in product development and increased development costs.
Challenge
The Denso EPS team is addressing key challenges in their hard real-time embedded software development, including:
- Unveiling Anomaly Detection for Performance Issues – Identifying and mitigating performance bottlenecks in Hard Real-Time EPS software to improve system reliability.
- Reducing the Number of Reviewers for Test Results – The increasing number of QA engineers is driving up costs and resource constraints, making it difficult to scale manual review processes.
- Enhancing Code Coverage Visibility – Improving insights into test and code coverage to elevate overall software quality and compliance with stringent automotive safety standards.
Aurora Labs Solution & Implementation
Denso deployed LOCI, Aurora Labs’ Lines of Code Intelligence platform, to address the growing challenges of quality and reliability in their mission-critical EPS system. Key aspects of the solution include:
- Vertical LLM for Performance, Reliability, and Quality Tasks – Aurora Labs deployed a domain-specific AI model focused on predicting and uncovering software performance and reliability issues in EPS.
- Predictive Anomaly Detection – LOCI enables early detection of failures in the development pipeline, significantly improving debugging efficiency.
- Automated Test Result Analysis – Reducing reliance on manual QA review efforts, leading to faster development cycles and lower operational costs.
- VPC-Based Deployment – Aurora Labs successfully deployed the solution within AWS VPC, introducing cloud-based observability and automation to a department that had never used cloud services before.
Results & Business Impact
Since deploying LOCI, Denso has observed:
- Early Detection of Performance Issues – LOCI’s anomaly detection enables proactive issue resolution.
- Efficient Resource Allocation – Automation has reduced the need for manual QA review, optimizing both budget and talent allocation.
- Increased Code Quality & Compliance – Improved visibility into test coverage has enhanced overall software reliability.
Next Steps & Opportunities
Denso’s EPS team is now expanding its adoption of LOCI and AWS, with key initiatives including:
- Scaling AI/ML-based automation with SageMaker to further optimize anomaly detection.
- Expanding LOCI for broader deployment and scaling within AWS VPC to enhance mission-critical system quality.
- Deepening cloud integration as this marks their first time directly paying for AWS services, with our team driving their cloud adoption strategy.
- Potential Expansion to Additional Departments – A successful continuation of this project in EPS will open the door for three additional Denso departments to adopt LOCI’s AWS VPC-based solution in 2025.