WildTrack: A Cloud-Native Wildlife Telemetry Data Analysis Platform Unlocking Migration Pattern Insights with AI

WildTrack is a cloud-native full-stack application that transforms raw wildlife telemetry data from Movebank.org into actionable conservation insights. It combines Java virtual threads, PostGIS spatial analysis, and generative AI to provide researchers with a real-time window into global migration patterns. Key technologies include AWS, Terraform, CI/CD, and Spring AI integration with Claude AI. The project aims to solve data overload issues for wildlife researchers by automating analysis workflows.

Wildlife researchers face data overload from Movebank's millions of GPS data points. Extracting valuable insights (e.g., weather-induced route deviations, animal entry into protected areas) is manual and time-consuming. WildTrack was created to automate this process, converting raw data into real-time, actionable insights for global migration pattern analysis.

1. High-performance Data Ingestion: Uses Java 21 virtual threads for high-concurrency batch pipelines from Movebank, with error handling for API issues (429/502 responses) and data validation (empty field detection, ingestion stats).
2. Spatial Data Analysis: Integrates PostGIS for boundary box queries, radius searches, and geofence alerts (email notifications on animal count changes).
3. Generative AI Query: Integrates Claude AI to convert natural language queries into PostGIS spatial queries (e.g., "Show sightings near British Virgin Islands in 2015" → spatial query).
4. Production Infrastructure: AWS ECS Fargate (serverless containers), RDS PostgreSQL (with PostGIS), ALB, Secrets Manager, Terraform (IaC), and GitHub Actions CI/CD.
5. Security: Input净化 (HTML stripping, length limits), RFC7807 error responses, field-level validation, and layered security groups.

WildTrack serves key use cases:

• Conservation Research: Analyze frigatebird GPS data (2014-2016) in the Caribbean.
• Migration Pattern Analysis: Identify weather-induced route deviations.
• Protected Area Management: Monitor animal entry into geofenced zones.
• Real-time Alerts: Notify researchers of animal count changes in specific areas.
• Data Democratization: Natural language queries lower technical barriers for non-technical researchers.

Cloud Architecture: Internet → ALB (public subnet, Shield Standard) → ECS Fargate Task (private subnet, pulls image from ECR, gets secrets from Secrets Manager) → RDS PostgreSQL + PostGIS (private subnet). Local Development: Use Docker Compose for DB, run Spring Boot app, explore API via Swagger UI.

WildTrack follows modern practices:

• Config Externalization: All env-driven config, no hardcoded secrets.
• Read-only Transactions: Minimize lock competition with @Transactional(readOnly=true).
• Pagination: All list endpoints support pagination to avoid unbounded queries.
• Secure Docker Builds: Non-root user, no source code in final image.
• Test Pyramid: Unit, slice, and integration tests ensure code quality.

WildTrack is an excellent example of applying modern tech to conservation. It solves real-world data overload issues and provides a complete production-grade solution. Key takeaways:

1. Virtual threads are effective for high-concurrency data ingestion.
2. PostGIS enables powerful spatial analysis for wildlife tracking.
3. Generative AI lowers barriers to data access for non-technical users. 4.IaC (Terraform) and CI/CD (GitHub Actions) ensure reproducible, scalable deployment. The project demonstrates how tech innovation can support conservation biology.