Submitted
2025 Global Health Challenge

Water Vision

Team Leader
Roshan Taneja
Water Vision combines machine learning with high-resolution satellite imagery to identify ideal locations for rainwater harvesting in underserved communities. We’ve trained a custom object detection model on 10,000+ annotated dwellings to map human habitation across 260 square miles of Northern Tanzania. This geospatial analysis combines on-ground interviews, participatory design, and local meteorological data to plan and deploy water harvesting units....
What is the name of your organization?
Maji Wells
What is the name of your solution?
Water Vision
Provide a one-line summary or tagline for your solution.
Improve access to clean water for 300M Africans using geo-spatial data and AI—to place rainwater solutions local to community and environment.
In what city, town, or region is your solution team headquartered?
Cupertino, CA, US and Losimingori, Tanzania
In what country is your solution team headquartered?
TZA
What type of organization is your solution team?
Nonprofit
Film your elevator pitch.
What specific problem are you solving?
The Maasai in Northern Tanzania walk 9-12 hours daily to collect water, a crisis worsened by climate change. The water they collect is unhygienic, resulting in illness. This reduces school attendance, increases health issues, and limits economic opportunity—particularly for women and children. Globally, over 2 billion people lack access to safe water. Traditional top-down interventions often fail to address local needs or optimize placement. Our solution targets these gaps by providing access to clean water using the best natural resources available [rainfall, rivers, etc.]. These deployments improve health and reduce illness, specifically for women and children.
What is your solution?
Water Vision combines machine learning with high-resolution satellite imagery to identify ideal locations for rainwater harvesting in underserved communities. We’ve trained a custom object detection model on 10,000+ annotated dwellings to map human habitation across 260 square miles of Northern Tanzania. This geospatial analysis combines on-ground interviews, participatory design, and local meteorological data to plan and deploy water harvesting units. Over 120 units have been deployed utilizing this technology. This approach was also awarded the NeurIPS 2024 Award for “ML for Social Impact” and has been presented at multiple academic conferences [NeurIPS 2024 and Machine Learning for Earth Observation 2024 Univ. of Exeter]. We want to scale up this problem across the board, modularizing the technology for as much of eastern Africa and other parts of the world as possible.
Who does your solution serve, and in what ways will the solution impact their lives?
We directly serve Indigenous Maasai communities in Tanzania, especially women and children burdened by long water collection journeys. Our solution reduces travel time by up to 7 hours/day, doubles school enrollment, improves health outcomes, and boosts local economic activity. By combining remote sensing with local knowledge, Water Vision ensures communities co-create solutions that are both scalable and rooted in trust.
Solution Team:
Roshan Taneja
Roshan Taneja
Co-Leader