Projects

Open-source smart bassinet for continuous neonatal monitoring in low-resource clinics. Targets the 2.4M annual neonatal deaths caused by undetected apnea, bradycardia, and hypoxia in facilities with no NICU equipment.

• —Dual-core architecture: bare-metal STM32 Cortex-M33 for real-time sensor acquisition (<5ms alarm latency), Cortex-A53 Linux for AI inference and clinical dashboard
• —3-layer false alarm elimination: dual-channel HR cross-validation (optical + electrical), multi-sensor apnea detection (MPU-6050 + camera), Edge Impulse anomaly classifier
• —Sensors: MAX30102 (SpO2/HR), ADS1115 + ECG pads, MLX90614 (non-contact temp), MPU-6050 (breathing via vibration), BME280, MIPI-CSI camera
• —On-device TTS voice alerts via HuggingFace — no internet required
• —Full App Lab clinical dashboard with Wi-Fi hotspot mode
• —Open-source BOM ~$150 vs $15,000–$50,000 commercial NICU monitors