Sneha Kulkarni Email & Phone Number

The Colony, Tx, Us

Mason, Ohio, United States

•Collaborated with business stakeholders to gather and document requirements for change requests raised by various departments, ensuring alignment with business objectives.•Performed impact analysis to evaluate the effects of change requests on existing systems, processes, and data flows, mitigating potential risks.•Wrote and executed complex SQL queries for data validation in System Integration Testing (SIT) and User Acceptance Testing (UAT) environments to ensure data accuracy and system integrity.•Managed and scheduled query execution in testing environments, ensuring timely validation and supporting accurate data transformation and migration processes.•Created detailed data mapping documents to define source-to-target data flow requirements, ensuring seamless data migration and transformation across systems.•Maintained comprehensive documentation of data validation processes and results, ensuring future reference for audits and compliance.•Collaborated with the Data Warehouse team to define Service Level Agreements (SLAs) for new change requests and proactively address data quality issues, enhancing overall data integrity and reliability.

Akron, Ohio, United States

•Consolidated contract data from various sources i.e., ERP & software portals using Google Cloud FileStorage and BigQuery. Automated data pipelines identified expiring contracts and spending trends, which led to an 8% reduction in contract renewal costs.•Utilized AWS (S3, Athena, EC2) to access and analyze facility supply data from invoices. Employed K-Means clustering to categorize spending and identify cost-saving opportunities. Reduced facility supply costs by 12% with the data dashboard built using PowerBI.•Built a custom Kaplan-Meier model from scratch on SageMaker using library scikit-learn to predict claim fraud over time. A serverless data pipeline utilizing AWS Step Functions & Lambda to extract claim data and feeding it to the model. This identified high-risk claims early, leading to a 21% reduction in fraudulent payouts.

● Calibrated a real-time patient sentiment feedback analysis pipeline, utilizing Apache Kafka for high-throughput data ingestionand Azure Cognitive Services for pre-trained NLP. Processed sentiment classification on over 2 million patient records, yieldiactionable insights that boosted customer satisfaction scores by 20% across CVS Caremark and Aetna.● Optimized an Azure-based patient churn prediction system using Azure Event Hubs for real-time ingestion and ApacheHadoop for distributed processing of 10TB of transaction and engagement data. The proactive interventions based on thesepredictions led to a 12% reduction in churn rate and an 18% increase in customer retention.● Developed a scalable patient analytics pipeline on Azure Databricks, utilizing Delta Lake for data consistency across billions ofrecords. Integrated MLflow for streamlined model tracking and periodic retraining, which improved prediction accuracy by 1and enabled timely, data-driven patient engagement strategies.● Built a medication adherence prediction platform using Apache Airflow for data orchestration and PyTorch for training a LoShort-Term Memory (LSTM) within a recurrent neural network model on more than 100 million records. This led to improvedmedication adherence through targeted conversions.● Augmented the customer-product knowledge graph with Spark SQL and GraphX and implemented a graph collaborative filterialgorithm Node2Vec to increase the recommendations, boosting customer engagement and sales.● Worked with COBOL and JCL to process legacy data for machine learning models and integrated them into the modern analyticsplatform, enhancing data accessibility and accuracy.

Pittsburgh, Pennsylvania, United States

•Built an Apache Spark ETL pipeline for legal/property data of 800GB. Extracted entities with SpaCy's NER and developed a recommendation engine Matrix Factorization using Spark MLlib, achieving high accuracy, F1-Score - 0.83. This reduced default resolution time by 24%.•Leveraged Azure Synapse Analytics and clustering algorithms - DBSCAN to segment mortgage servicer data, uncovering hidden patterns that led to tailored subservicing strategies. This approach improved an overall operational efficiency by 30%.•Preprocessed customer data with Pandas & Scikit-learn and developed a Cox Proportional Hazards model for CLV prediction, yielding high predictive ratios of 0.74 Harrell's C-index, resulting in building potent customer retention strategies.

India

•Employed Gaussian Process regression, a non-parametric Bayesian approach, to model the anisotropic trends within wellbore data, improving image quality in depth migration for accurate reservoir characterization.•Implemented a hybrid demand forecasting model with a team of 3, combining ARIMA - Autoregressive Integrated Moving Average for capturing seasonality and XGBoost for non-linearity patterns within sales data by utilizing Pandas & Matplotlib for data manipulation and visualization. This strategy achieved high forecasting accuracy with Mean Absolute Percentage Error < 10%.•Designed and implemented an automated ETL pipeline by utilizing Python framework Kedro, ensuring timely data availability for report generation. Integrated Tableau with python script to create dynamic & interactive reports giving insights into crucial KPIs

India

•Collaborated with the client’s supply chain team to identify demand drivers. Built an LightGBM model on SAP S/4HANA data for demand forecasting. Implemented statistical safety stock model - Wilson EOQ for inventory optimization.•Built end-to-end ML model using AWS SageMaker to automatically classify procurement spend data into predefined categories. This model utilized an algorithm - Support Vector Machine (SVM) to categorize spend based on historical data patterns, resulting in a 12% reduction in overall procurement spend.

Masters, Computer Science, A

Coursework : Data Mining, Database Systems, Artificial Intelligence, Cloud Computing, Computer , Network & Security, Advanced Operating.

Bachelor Of Engineering - Be, Computer Science