Process and match
Python pipelines ingest and validate sample road data. A C++ matching component connects road and sign observations.
The engineering challenge
A value is only as useful as the evidence behind it.
Road data can be incomplete, inconsistent or stale. GeoSpeed brings evidence, baseline inference and quality review into one workflow: what speed limit is supported, how confident is the result, and what still needs attention?
A system across disciplines
Data. Intelligence. Delivery.
Infer and explain
A FastAPI service supports baseline speed-limit inference and confidence evaluation, with evidence and conflicts exposed for review.
Serve and review
Java Spring Boot APIs and React / MapLibre interfaces connect data products to map-quality and partner-review workflows.
Beyond a standalone model
Engineering the surrounding workflow
Automotive integration demonstration
An in-vehicle simulator supports route replay, speed-limit alerts, simulated vehicle signals and driver-assistance mismatch scenarios. Partner issue triage and launch-readiness workflows make integration concerns visible.
Quality before release
Release rules consider confidence, freshness, conflicting evidence and unresolved high-severity issues. Observed vehicle speeds are validation signals, never substitutes for legal speed limits.
Software delivery foundations
Docker Compose, automated tests and continuous integration support repeatable development across Python, Java, C++ and TypeScript components.
Demonstration scope
The current project uses small sample datasets and simulated signals. It demonstrates an integrated architecture; it is not a claim of a deployed navigation product, validated vehicle safety or full jurisdiction-scale data coverage.