Team - Payments Data & AnalyticsTech stack - Pyspark, Python, SQL, Schell scripting, AWS, Databricks● Developed a model to identify non-client corporate entities that our clients transact with by using third-party data comprising of 100 mil records. Trained an optimized random forest regressor on string distance features based off entity names coming from the 2 sources. Resolved a sample of high-value non-client corporate entities with 89% accuracy.● Built a pipeline to read and flatten raw data nested upto 14 levels deep, and store it in tabular format in spark to be readily ingested for analytics use cases.

Subject - Data Mining and Predictive Analytics in R

Team - Uber busTech stack - Presto, Python, Kepler.gl● Quantified performance of bus routes by developing 17 KPIs and ingesting them to build a route ranking model. Based on results, suggested improvements in marketing strategy.● Segmented riders into 3 engagement buckets by analyzing rider behavior. Assessed the impact of price-hike by identifying churn rate among all riders.● Created a dashboard to exhibit product performance through route visualizations and KPIs.

Tech Stack: Python, MySQL, Tableau● Developed a recommender system using content-based filtering for suggesting books to readers.● Conducted a comparative analysis on the popularity trend of New York Times bestseller books vis-à-vis other books.● Analyzed reading behavior of ~20k library patrons for e-books and audiobooks.

Team - Machine Learning Centre of ExcellenceTech stack - Python, Pyspark, SQL, Hive, Impala, Amazon Web Services, Tableau, Teradata● Identified target customers using k-means clustering followed by regression.● Identified loan bearers who are most likely to default on monthly repayments using machine learning.● Automated the manual method of mapping merchants to brands by computing sequential similarity between merchants and brand names.● Recognized prospective customers for mortgage that dropped off at different stages of online application using data captured by Adobe SiteCatalyst. Recommended UI changes to improve customer experience.● Built an interactive voice service using AWS to respond to users with insights about spending patterns of UK population.

Team - Enterprise ArchitectureTech stack - Kafka, Java, HTML/CSS● Benchmarked a publish-subscribe messaging system rooted on Kafka clusters by optimizing Kafka configurations.

● Deep dive into Reliance Jio delivery process framework