Learning eoAPI – Deploying Titiler and TiPG on Replit

[nishadhka]

Thanks to the great suggestion to learn EoAPI by deploying its components outside the typical Kubernetes and Docker-based setup. That was a fantastic idea — exploring a simpler path to understanding the ecosystem.

The Replit templates shared below offer an excellent starting point, using its in-house postgresql and easy to deployment facility, backed by AI Agent. The .replit file in the template shows the server setup and the main Python file being deployed, along with documentation (.md) files and other Python files written by the AI agent.

• [Titiler + PgSTAC Template](https://replit.com/@e4drr/TitilerPgstac-template#.replit)
• [TiPG + Tiler Template](https://replit.com/@e4drr/TipgTiler-template#.replit)

These templates use Replit’s built-in PostgreSQL and AI agent support, making it easy to experiment with Titiler (for serving raster tiles from COGs) and TiPG (for vector tile services). This no-Kubernetes approach is especially helpful for learners and quick prototyping. It also helps overcome common deployment issues with Titiler, such as setting up a reverse proxy, and makes it easier to understand TiPG's use of vector tiles (MVT) with help from the AI agent.

Example applications:

• COG Visualization with Titiler: A hands-on Jupyter notebook example using Replit deployed titiler endpoint
  [Titiler Tiles COG – Replit Notebook](https://github.com/icpac-igad/arco-ibf/blob/titiler-pgstac/titiler-tiles-COG-replit.ipynb)

• Vector Tiles with TiPG: Visualizing river networks using vector tile services — on working with mvt formats.
  https://github.com/icpac-igad/arco-ibf/tree/tipg/rivers and using it for frontend https://github.com/icpac-igad/arco-ibf/blob/155dfb3115ccc329f45a3ec43b4d833c62072b75/frontend/river_network_visualization.html#L296

These explorations would help in pivot into k8s setup and or using a serverless setup, such as GCP Cloud Run, to deploy Titiler and TiPG, and to explore other components of eoAPI. Hopefully, the serverless setup can still support the integration of EoAPI’s interlinked components originally designed for Kubernetes.