Natural disasters like floods and earthquakes necessitate efficient early warning systems to mitigate risks. This paper proposes an IoT-based disaster management system integrating ESP32 microcontrollers, environmental sensors (water level, rainfall, seismic), and machine learning (ID3 algorithm) for real-time flood and earthquake detection. The system achieves 95% flood detection accuracy (3% false positives) and 92% earthquake detection accuracy (5% false positives), with alerts delivered via Telegram bot and web dashboard within 8–10 s. Live weather API data enhances prediction, while modular architecture ensures scalability. Testing demonstrates 99.1% data transmission reliability and 90% user satisfaction, offering a cost-effective, resilient solution for disaster-prone regions. Limitations include power dependency, addressed through future solar integration.