 This capstone challenges students to transform a passive solar-powered water filtration unit into an intelligent, self-monitoring global health node powered by Agentic AI. Teams will integrate three AI systems into a single deployed platform, targeting communities in Cuba and South Africa currently relying on static humanitarian water infrastructure.
Pillar 1 — Real-Time Pathogen Detection Students integrate computer vision and biophotonic deep learning sensors to scan water at the molecular level, identifying pathogens such as Vibrio cholerae and E. coli in minutes. The system automatically pushes SMS health alerts to local community coordinators before any water is consumed, collapsing outbreak response time from days to minutes. Pillar 2 — Predictive Maintenance via Digital Twin Each deployed unit is mirrored by a cloud-hosted Digital Twin that ingests real-time sensor telemetry including flow rate, pressure differentials, and TDS levels. A predictive ML model forecasts component failure before it occurs and automatically routes a service request to the foundation’s logistics team, ensuring units never go offline unattended. Pillar 3 — Geospatial Solar Optimization A generative AI model analyzes satellite imagery and real-time weather data specific to each unit’s GPS coordinates. The system continuously adjusts solar panel angle and filtration throughput speed to maximize clean water output based on actual solar irradiance conditions at that specific deployment site. |
Required Student Skills |
| Machine learning and deep learning (model training, inference pipelines, CNNs)
Python programming (NumPy, pandas, TensorFlow or PyTorch) Embedded systems and sensor integration (IoT hardware, serial/I2C/SPI protocols) REST API development and cloud platform experience (AWS, Azure, or GCP) Version control and collaborative software development (Git/GitHub) |
Recommended Student Skills |
| Computer vision and image processing (OpenCV, biophotonics sensor pipelines)
Time-series data analysis and anomaly detection (InfluxDB, LSTM or XGBoost models) Geospatial data and mapping APIs (Mapbox, Google Solar API, satellite imagery processing) Dashboard and data visualization development (Grafana, React/Next.js) Digital Twin modeling (Azure Digital Twins or AWS IoT TwinMaker) |
| Hardware/Software Requirements |
| Hardware: NVIDIA Jetson Orin (edge AI compute), Raspberry Pi HQ Camera with spectral filters, flow rate sensor, pressure differential sensor, TDS (Total Dissolved Solids) sensor, servo motors for solar panel actuation, solar panel array with motor driver
Software / Cloud: Python 3.x, TensorFlow or PyTorch, TensorFlow Lite (edge inference), OpenCV, InfluxDB (time-series telemetry), Azure Digital Twins or AWS IoT TwinMaker, Twilio SMS API, Open-Meteo or Tomorrow.io weather API, Google Solar API, Grafana or React/Next.js dashboard, Git/GitHub Sponsor-Provided: Access to existing solar-powered filtration unit hardware and deployment site GPS coordinates for Cuba and South Africa locations |
Leave a Reply
Your email is safe with us.