Hivemapper is building a decentralized street-level mapping network that collects dashcam imagery from a global community of drivers and uses computer vision and machine learning to produce continuously updated maps. The platform spans six continents, with contributors capturing imagery every day and engineers processing millions of kilometres of that data weekly through distributed systems. Contributors - both individual drivers and fleet operators - install dashcams and collect data passively as they drive, earning HONEY tokens in return. The model positions contributors as owners of the network rather than passive data suppliers.
The technical stack is broad and demanding. Engineers work across computer vision, sensor fusion, geospatial processing, edge computing, and cluster-scale distributed systems. AI models are trained and validated with input from the contributor network itself, creating a human-in-the-loop pipeline that operates at significant scale. The platform also exposes APIs for third parties to consume map data and build on the underlying infrastructure, which is designed to be permissionless and transparent.
Hivemapper's core technical domains include:
- Computer vision and machine learning for imagery processing and map generation
- Distributed and edge computing systems handling high-throughput data pipelines
- Geospatial and sensor fusion for accurate, fresh street-level mapping
- Token-based incentive systems and decentralised physical infrastructure networks
- API development for map data consumption and third-party integration
The engineering culture is described internally as fast-moving, with an emphasis on shipping production-grade machine learning solutions that run both at the edge and in large computing clusters. The company frames its approach as a departure from traditional mapping - replacing centralised, infrequently updated datasets with a contributor-driven network capable of refreshing street-level imagery on a daily basis across a global footprint.