Tiyao Li | Senior Software Developer — Backend | Cloud | Data & AI Platform | AI Infra

End-to-end ownership of a cloud-native auto-labeling Data & AI Platform for autonomous-driving perception data, covering architecture, service APIs, data pipelines, async processing, deployment, observability, and operations, with 140TB+ processed across production workflows.

• Designed, built, and operated a cloud-native auto-labeling Data & AI Platform for autonomous-driving perception data across LiDAR and camera workflows, with a Python/FastAPI backend service handling job orchestration, request validation, workflow control, and reliable production execution.
• Integrated deep-learning inference workflows for 2D/3D perception data into auto-labeling pipelines, and built an AI agent for automated job log analysis, root cause diagnosis, and remediation suggestions.
• Built event-driven processing with Kafka and KEDA-based autoscaling, using MongoDB for job metadata, configuration, processing state, and labeling results.
• Provisioned and managed cloud-native infrastructure using Terraform, including Azure services such as Blob Storage, Azure Container Registry, and Key Vault, and supported Kubernetes workloads on Red Hat OpenShift clusters.
• Diagnosed and resolved production issues across OpenShift/Kubernetes, Kafka, application logs, cloud configuration, and service reliability, improving platform stability and operational efficiency.
• Worked directly with internal users and stakeholders to clarify requirements, investigate issues, and deliver practical platform improvements.

Developed software-driven validation toolchains for radar-based ADAS truck systems, enabling repeatable sensor data acquisition, signal monitoring, and automated measurement workflows across HIL and in-vehicle test environments.

• Integrated radar sensors, vehicle signals, measurement tools, and validation software into HIL environments for Daimler and Volvo truck radar validation projects.
• Developed CANape-based measurement and monitoring setups for radar, LiDAR, camera, and vehicle-signal data across bench, in-vehicle, and global endurance-run scenarios.
• Built automation workflows for sensor data acquisition, online monitoring, signal plausibility checks, and validation processes, reducing manual setup effort.
• Enabled 100,000+ km of global truck endurance-run data acquisition across Japan, Europe, the United States, and China through CANape-based measurement setups and configuration packages for long-duration sensor and vehicle-signal data aggregation.
• Improved measurement reliability through data consistency reviews, standardized validation procedures, and reproducible measurement workflows.
• Contributed to Jenkins-based automation pipelines to improve workflow reproducibility and streamline measurement execution.