• Preprocessed data, including embedding vectors and original text chunks, to facilitate downstream clustering and labeling tasks. Managed dimensionality reduction using PCA, t-SNE, or UMAP.• Implemented KMeans and Agglomerative Clustering algorithms, tuning hyperparameters to accurately segment text data into distinct clusters. Calculated and visualized cluster centers to support interpretability in reduced dimensionality spaces.• Designed a process to generate concise semantic labels for each cluster by synthesizing text chunks and querying OpenAI's language model API. Developed logic to batch text input, reducing API calls while ensuring labels are meaningful and informative.• Created interactive visualizations of clusters in 2D and 3D, adding descriptive labels for each cluster. Integrated interactive CheckButtons, allowing users to toggle cluster visibility for enhanced exploratory analysis and better insights into cluster distributions and relationships.

• Spearheaded a team of three at Optum to perform migration of a complex AbInitio data pipeline to Azure cloud.• Modernized on-premise data system by moving to Azure cloud, reducing cost and improving data security.• Migrated AbInitio pipelines to Python-based system for healthcare rebates data, cutting operational cost by 30%.• Developed and optimized Azure Databricks pipelines, decreasing data processing time by 35% and increasing data availability for downstream analytics in pharmaceutical claims operations.• Mentored data engineers to identify and resolve performance bottlenecks within pipelines, leading to a 50% reduction in processing time and increased productivity.

• Served as offshore tech-lead at Capgemini for providing technical guidance to off-shore team, and ensuring smooth collaboration with on-site colleagues.• Conceptualized a non-prod environment on production cluster and collaborated with teammates to integrate under-development projects ensuring 50% lesser time and efforts while release to production.• Created a Scala-based auditing process to keep a track of vital stats for data application’s daily run.• Designed and implemented a disaster-recovery process which enable applications to run on an alternate server to safeguard critical information and minimize downtime during unforeseen events.

• Collaborated with cross-functional teams, data scientists, and analysts to understand business data requirements and developed end-to-end ETL solutions in an agile methodology setup using big data tools with Python & SQL.• Leveraged Python to migrate pre-existing SAS coded data pipelines to spark framework that reduced cost by 25%.• Assumed responsibility for data quality assurance tasks to address sudden unavailability of the testing resource, preventing a one-month operational delay.