We present the implementation of seizure detection algorithms based on a minimal number of EEG channels on a parallel ultra-low-power embedded platform. The analyses are based on the CHB-MIT dataset, and include explorations of different classification approaches and different pre/post-processing techniques to maximize sensitivity while guaranteeing no false alarms. For 8s window size and subject-specific approach, we report zero false positives and 100% sensitivity. These algorithms are parallelized and optimized for a parallel ultra-low-power platform, enabling 300h of continuous monitoring on a 300 mAh battery, in a wearable form factor and power budget.