Sriram Uppula

• Developed and deployed statistical and machine learning models using Python, SQL, and Scikit-learn to support risk, fraud, and customer analytics initiatives.
• Engineered and prepared large datasets by performing data cleaning, feature extraction, outlier treatment, and variable transformations, improving model accuracy by 20%.
• Built automated data pipelines for model training and scoring, integrating structured and unstructured data from enterprise platforms, reducing manual processes by 30%.
• Conducted exploratory data analysis (EDA) to identify trends, anomalies, and key drivers impacting business KPIs, delivering insights to product, credit, and operations teams.
  • Designed and implemented model performance dashboards in Tableau and internal BI tools, enabling leadership to track drift, accuracy, and operational performance in real time.
  • Collaborated with data engineers to optimize data ingestion layers and ensure the availability of high-quality, analytics-ready datasets for modeling workloads.
• Applied statistical techniques such as regression, classification, hypothesis testing, and clustering to support business decision-making.
• Supported production ML workflows by monitoring model drift, retraining cycles, and data quality signals, ensuring consistent model performance and compliance with governance standards.
• Documented model development, validation, and deployment processes to meet Truist’s internal audit, compliance, and model-risk management requirements.

• Designed and engineered large-scale ETL/ELT pipelines using PySpark, Hadoop, Hive, and Python, increasing overall data processing efficiency by 30%.
  • Automated data quality frameworks (schema checks, null thresholds, business rule validation), improving data reliability, and reducing manual QA by 40%.
  • Optimized SQL/Hive queries and Spark jobs through partitioning, caching, and tuning techniques, reducing query execution time by 45%.
• Developed ingestion workflows for APIs, flat files, and RDBMS sources, and integrated them into AWS Redshift and GCP BigQuery cloud data warehouses.
• Built and maintained Airflow DAGs for batch/stream pipelines, improving pipeline stability and reducing operational failures by 35%.
  • Implemented logging, monitoring, and alerting using CloudWatch/Stackdriver, improving observability, and achieving 99% pipeline uptime.
• Collaborated closely with data scientists and analysts to supply feature-ready and analytics-ready datasets for modeling, dashboarding, and KPI reporting.
• Created reusable PySpark libraries for transformations, normalization, and deduplication, reducing new-source onboarding time by 25%.
• Performed data modeling and applied partitioning/bucketing strategies to optimize storage and accelerate analytical workloads.

• Conducted RCA and production support for ETL failures, implementing fixes that reduced recurring issues by 50%.

• Improved operational reporting speed by 35% by automating SQL pipelines and creating reusable data transformation scripts.
  • Conducted in-depth EDA on customer, seller, and logistics datasets, uncovering trends that reduced delivery delays by 12%.
  • Built interactive dashboards using Tableau and Power BI, increasing stakeholder visibility into KPIs, and improving decision-making efficiency by 30%.
• Integrated data from MySQL, Hive, and API sources to create unified analytical datasets, reducing data preparation time by 25%.
  • Implemented statistical techniques (forecasting, regression, clustering) to support pricing, inventory management, and marketplace operations.
• Identified root causes behind shipment delays, seller performance dips, and order cancellations, reducing recurring operational issues by 15%.
• Partnered with business teams to convert complex data into actionable recommendations that drove measurable improvements in customer experience.