Routescanner

The Challenge

Routescanner helps freight forwarders and carriers find the fastest, cheapest container routes across the globe. The platform ingests millions of sailing schedules from dozens of carriers, each with its own format, quality level, and update cadence. Route calculations depend on a GIS environment that must resolve complex multi-leg geographic queries in real time. The previous architecture was hitting performance ceilings, and onboarding new data sources took weeks of manual effort.

My Contribution

Took ownership of the full stack: React/Remix frontend, Node.js backend services, and the PostgreSQL/PostGIS data layer. Redesigned the GIS pipeline around custom geo-search algorithms, cutting average route query times from seconds to under 200ms.

Built automated ETL pipelines that normalize, validate, and deduplicate carrier schedules on ingestion, reducing new-source onboarding from weeks to days. Introduced structured code review, TypeScript strict mode across services, and mentored three junior developers into independent contributors.

In the platform's most recent phase, as Routescanner transitioned to its new owner, I took ownership of the infrastructure and developer platform. I codified the AWS/ECS environment in Terraform, hardened the GitHub Actions CI/CD pipeline with branch protection and tighter security settings, and migrated this and sister repositories off paid Mapbox onto open-source Leaflet to remove licensing cost and vendor lock-in. I also built the AI sandbox the new owner asked for: a one-to-one copy of staging where non-developers describe a feature in plain language, an LLM prototypes it on a throwaway branch, and developers take it production-ready after review and testing — evolving an earlier branch-per-prototype deploy flow into a safer, guard-railed setup.

Key Results

• Route query p95 latency dropped from ~4s to under 200ms after GIS redesign
• New carrier data source onboarding reduced from weeks to 2-3 days
• Frontend Lighthouse scores improved to 90+ across all key pages
• Established code review culture and TypeScript strict mode across the codebase
• Owned the platform end-to-end: AWS/ECS in Terraform, hardened GitHub Actions CI/CD and branch protection, and a Mapbox-to-Leaflet migration across repos that cut licensing cost and vendor lock-in
• Built the AI sandbox the new owner requested: a staging clone where non-developers prototype features with an LLM on isolated branches, which developers take production-ready after review

Technologies Used