Building a geospatial platform from zero

The starting point

I joined Aereo as one of the founding engineering members. There was no platform yet: no frontend architecture, no processing pipeline, no deployment story. Over the following four years I helped build all three, and the architecture decisions made in those first months are still the ones the team works inside today.

What I built in the first year

• Foundational frontend architecture, later adopted as the team-wide engineering standard.
• The team's first private npm module (@aus-platform/cesium) for scalable geospatial frontend development.
• Terrain tile generation systems, COG generators, MBTiles extraction workflows, and vector tile servers, benchmarking EC2 instance types for the performance-per-cost tradeoff rather than guessing.
• Foundational event-driven AWS Batch job orchestration, and an S3-to-S3 export Lambda enabling ORI/DSM data portability across processing and analytics platforms.

Then making it survive real customers

Building v1 is the easy half. The platform then had to hold up under enterprise procurement, government data residency rules, and customers whose sites had no reliable internet. I single-handedly managed infrastructure communication and the v1 rollout for a major steel enterprise, implementing self-hosted geospatial delivery that satisfied their data residency requirements, a deployment that would otherwise have been blocked outright.

Shipping a second product on top

In 2024 I led and shipped Aereo Go v1, the company's first dedicated mobile product, coordinating backend, mobile, and DevOps teams from inception through production launch. Same pattern as the platform itself: no existing product, a cross-functional team, and a launch date.

Why this matters if you're hiring

Most engineers you can hire have worked inside a product that already existed. If yours doesn't exist yet, or exists as a prototype that won't survive its first real customer, the relevant question is whether someone has carried that specific weight before: choosing the architecture, building it, and then being there when enterprise customers stress it. I have, once, all the way through.