Night Watch, a winning hackathon project from King Mongkut's Institute of Technology Ladkrabang, uses satellite imagery (VIIRS) and AI to detect failed streetlights in Thailand. By analyzing nighttime satellite data alongside GIS records and weather patterns, the system achieves 90.1% accuracy and reduces required inspection patrols from ~33/month to ~2.7/month—a 10x improvement. First-year cost is ฿280,000 (~0.49% of a district's annual lighting budget), making it highly actionable for government agencies seeking cost-efficient infrastructure maintenance.