Cross-platform engagement forecasting

At DosBro, I deployed an LSTM + XGBoost hybrid model to forecast cross-platform user engagement, giving product teams a signal for when and how to push notifications instead of blasting everyone on a fixed schedule.

The hybrid beat baseline methods by 15% on prediction accuracy and informed data-driven retention campaigns across the company’s digital platforms.

Focus

• Hybrid deep learning and gradient boosting for forecasting
• Engagement prediction across multiple platforms
• Product-facing ML for notification strategy