Kubernetes and AI To Protect Our Forests: A Cloud Native Infrastructure for Wildfire Prevention
Talkat KubeCon + CloudNativeCon Europe 2025 · on Friday, 2025-04-04· at 13:45· Level 0 | ICC Capital Hall | Room J·
summary
Startup moves from JupyterHub notebooks to a Kubernetes-based stack using Daxter + Kubernetes, GKE, Google Cloud Run, Pub/Sub, Grafana and BigQuery, plus GPU nodes and ephemeral volumes. Pipeline code lives in Python ops with an internal Maple library and runs on 15 cm satellite imagery combined with utility pole coordinates. Result: first end-to-end 30 minutes zero-touch delivery achieved; org scales from ~20 to over 100 people while reducing manual handoffs.
AI-written summary · about
chapters (17)
- 0:00Wildfire prevention with Kubernetes & AI
- 4:32Satellite imagery & ML for wildfire risks
- 6:39Kubernetes & Jupyter Hub at Overstory
- 8:22Jupyter notebooks & manual workflows
- 9:48Automating workflows with Dagster
- 11:13Dagster’s cloud-native architecture
- 13:19Dagster’s IO managers & pipelines
- 16:51Customizing Kubernetes with Maple
- 18:35Single-pod vs multi-pod pipelines
- 20:43Building orchestration platform
- 22:47Observability with Grafana & OpenTelemetry
- 24:13Zero-touch delivery in 30 min
- 25:17Future pipeline reliability & resources
- 26:41Q&A: New Dagster features
- 29:12Q&A: Custom platform outside Dagster
- 30:39Q&A: Resources in single vs multi-pod
- 31:47Q&A: Exploring Dagster Pipes
related talks
27mGen-AI at Scale: Simplifying Orchestration of Healthcare Applications Across Multi-Cluster Environme
19mScalable Data and AI Solutions with Cloud-Native Architecture
27mHarnessing Kubernetes for Scalable AI/ML Workloads
43mKeynote: Building AI in the Open: How the Cloud Native Community Shapes What Comes Next
54mEconomics and Best Practices of Running AI/ML Workloads on Kubernetes
30mDesigning a Hybrid AI Platform on Kubernetes
42mKubernetes for Machine Learning: Managing and Scaling AI Workloads
40mStreet fighting techniques for multi tenant machine learning...