I built the Cart Abandonment Intelligence System as a personal portfolio project, combining an XGBoost prediction pipeline with SHAP explainability and a RAG-based diagnosis pipeline. It achieved ~90% classification accuracy and generated personalized intervention strategies using GeminiAPI.
I also developed a multi-page Streamlit dashboard for prediction, explainability, diagnosis, and strategy generation. The project brought together ML prediction, what-if simulation, intervention impact estimation, and LLM-powered recommendations.
At SysCloud, I analyzed operational datasets using SQL and Excel, automated data-processing and reporting workflows with Python, and built Excel reports and dashboards. I also built a Power BI dashboard to explore Google Play Store app ratings, installs, reviews, pricing, and category trends.
