Tharun Karasani

● Leveraged MIP optimization with Gurobi to deploy a promotion and pricing strategy for a major retail brand, maximizing different business objectives ● Modeled sales volumes with 87% accuracy for hundreds of products across dozens of retailers in North America and LATAM using hierarchical mixed models ● Delivered a production-ready web app UI solution that drove significant sales growth and supported multi-year commercial targets ● Managed a team of five data scientists, delivering strategic insights to senior executives to enable data-informed decision-making ● Deployed automated machine learning models at scale, unlocking substantial incremental revenue opportunities ● Applied GenAI-driven predictions to identify new growth opportunities in supply chain and operations ● Optimized inventory, forecasted vehicle demand, and modeled dealership incentives for automobile brand across 100+ districts in the USA, achieving significant cost savings and wholesale improvements ● Applied histogram-based gradient boosting and mixed integer programming to optimize vehicle allocations with low forecast error (3% WMAPE)

● Assessed delivery risk of aircraft parts suppliers using multivariate ordinal time series prediction with Cumulative Link Models (CLMs) ● Leveraged Dynamic time warping for time series clustering and meta learning to obtain 92% accuracy to predict new supplier’s risk rank ● Onboarded three global Boeing teams to the risk assessment tool through training, documentation, and support

● Worked on customer personalization for a B2C retail client focusing on targeted marketing recommendations through customer segmentation and purchase propensity models ● Optimized PySpark scripts to reduce pipeline process time by 80% that models terabytes of clickstream, marketing & customer data ● Leveraged Databricks dbx framework to automate the cloud-agnostic deployment of pipelines that ingest data into Big Query & Snowflake

● Deployed machine learning models and performed prescriptive analytics to generate business insights for operational efficiency o Achieved 70% reduction in Trade Fail rate by identifying fail root causes using a Stacked generalization of K-prototypes clustering, Bagging, and FP-Growth pattern mining algorithms o Automated 30% of the manual workflow by resolving trade flow bottlenecks using unsupervised, distance models in settlements o Leveraged Local Outlier Factor (LOF) algorithms in PyCaret for anomaly detection in client trading patterns o Accelerated model process time by 1.5 times through dynamic multiprocessing and threading in the ML Pipeline o Enabled data-driven decision-making by publicizing powerful Tableau dashboards among business teams with relevant KPIs ● Performed data modeling, pipelining, and processing using PySpark, Alteryx and visualized it through live, interactive Tableau dashboards ● Improved ETL processing speed by 50% after pipelining data through optimized PySpark refiners and leveraging OLAP models ● Goldman Sachs COVID-19 Data-thon winner for proposing an effective vaccine distribution algorithm using ANN

• Analyzed in real-time the usage of underground water in a village using Machine Learning and Hadoop ecosystem • Deployed a system of sensors, and coded microcomputers to capture and transmit the real-time field data to the executor nodes • Monitored an average excess usage reduction of 200 gallons per family post prototype deployment • Leveraged regression models to predict the water resource availability based on the current consumption metrics • Automated the extraction, processing pipeline, and storage using SPARK and HDFS • Productionized the Tableau dashboard layer to monitor the availability and usage of water resources

• Designed and developed transactional and analytical data structures. • Worked closely with clients to establish problem specifications and system designs. • Wrote and implemented scripts to enhance user experience and integrated scripts. • Designed processing applications in HADOOP architecture. • Coded 2 websites using web development languages i.e. HTML, CSS, JAVASCRIPT, Angular JS, SQL.

• Worked on Algorithms in Network security and cryptography, Design for Testability and BUS Protocols using Xilinx and SPARTAN 3E XS3C1200E FPGA. • Implemented Industry related techniques for further improvement on projects. • Analysed Altera and Spartan 3E family internal FPGA architectures, configuring and understanding process flow.

Trained in VLSI based mini projects and sub-blocks design for the dynamic algorithm building.