At HSBC India, I engineered a Flask-based marketing campaign analytics platform that standardized lead tracking across more than 15 sources. It saved $45K annually by reducing reconciliation time and duplicate records, and I designed a campaign reference code system to improve traceability.
I developed Python, PySpark, and SQL pipelines for marketing data and Google Cloud BigQuery, and built Tableau dashboards for campaign managers. I also developed a PySpark and XGBoost wealth management churn and propensity model, achieving an 86% AUC and using SHAP values to explain risk drivers.
For financial disclosures, I built an ETL pipeline that transformed PDF and HTML documents into structured datasets, then added NLP sentiment indicators for trend analysis and risk reporting. In my FinSight project, I built a Python RAG pipeline for SEC filings with ChromaDB, sentence-transformer embeddings, Anthropic Claude, and a Streamlit interface.
