Bikumalla sahithi

•Built and operationalized fraud detection models in Python/Scikit-Learn on SageMaker, tuning features/thresholds and monitoring drift to lift fraud catch rate +18% with lower false positives, directly reducing claim leakage. •Automated ETL/ELT for policyholder, quote, and claims data with AWS Glue, Airflow, SQL, and data quality checks (row counts, schema, outlier flags), enabling T+1 refresh for analytics and cutting manual reporting effort −60%. •Delivered executive Tableau/Power BI dashboards for churn, LTV, and retention cohorts; standardized KPI definitions and embedded RLS, driving adoption across Product, CX, and Finance and shortening decision cycles from weeks to days. •Migrated legacy marts to AWS Redshift and optimized queries (sort/dist keys, materialized views), reducing key BI workloads −55% and saving infra cost via workload management and compression tuning. •Implemented forecasting (Prophet/ARIMA) and uplift models to inform pricing/renewal strategies; increased underwriting accuracy +12% and improved renewal targeting for high-value segments. •Instituted CI/CD for analytics (GitHub Actions, code review, unit tests for SQL/Python) so models and dashboards ship reliably with versioned artifacts and rollback paths.

•Conducted large-scale EDA on telecom, BFSI, and retail client datasets (customer churn, credit risk, fraud, claims), engineering features (usage patterns, payment behavior, geodemographics) that improved ML model AUC by +0.08 and reduced business losses −12%. •Designed and deployed machine learning pipelines (Logistic Regression, Random Forest, XGBoost) with cross-validation, hyperparameter tuning, and SHAP explainability to deliver predictive scores that cut manual reviews −30% and accelerated decision-making. •Developed Power BI executive dashboards (revenue leakage, churn cohorts, loan delinquency, sales pipeline) with drill-through, KPI alerts, and row-level security, enabling single-source-of-truth reporting across client business units. • Optimized SQL/ETL workflows (window functions, CTEs, incremental loads) to reduce reporting runtime from 4h → <40m, enforcing reconciliation logs and exception handling for high-availability analytics. •Automated regulatory and compliance reporting (RBI, IRDAI, CIBIL equivalents) using parameterized SQL, validation gates, and audit trails, reducing compliance effort by 55% while improving accuracy. •Partnered with cross-functional client teams (Finance, Risk, Marketing) to translate analytics into business actions (limit strategies, targeted campaigns, pricing adjustments), tracked via uplift against baselines.

• Cleaned, profiled, and standardized client datasets using Python (Pandas/NumPy) and SQL; implemented data quality checks (duplicates, nulls, business rules) that boosted model reliability and reporting accuracy. • Built prototype dashboards in Power BI and Excel for loan performance, customer churn, and sales productivity, later transitioned into production-ready BI solutions for clients. • Authored reusable SQL scripts and stored procedures for recurring operational and compliance reports; reduced ad-hoc reporting requests −20% and improved SLA adherence. • Assisted in fraud detection and churn prediction experiments, handling class imbalance (SMOTE, stratified sampling) and documenting evaluation frameworks to accelerate production model rollouts.