Rohith Upadhyayula

Daily flow plan for supply chain of a retail giant

• Enabled collaborative planning between workforce management, logistics and store operations teams by developing a daily flow plan for their retailer's distribution centers
• Designed and developed an end to end analytical pipeline to gather data from multiple teams, implemented forecasting models to predict flow volume for the upcoming 120 days, built intuitive visualizations to track future flow plan and understand risks and created an insights engine that generated key recommendations for work force planning and capacity planning at the retailer's distribution centers
• Used an ensemble of time series models (Prophet, SARIMA, Holt’s Winter) in python to develop a forecasting model that gave accuracies in excess of 94% at a daily level. A Web UI was built for visualizing the flow plan as well as the distribution centers which could be at risk of a capacity breach in the upcoming days

Retail sales forecasting and Promotion analysis of prominent video game retailer

• Worked with the FP&A team for predicting unit sales, dollar sales and transactions for brick and mortar channel and Online channel at weekly level across various product categories for inventory optimization and maximizing margin
• Implemented time series forecasting with FB Prophet added additional product category specific regressors along with holidays for model tuning and generating more accurate forecast
• Worked with the pricing and promotions team for measuring the performance of past promotions for online and in store channel to help them in promotion planning and marketing
• Power BI dashboard was built to track promotion performance based on lift analysis in terms of unit sales, dollar sales and margin

Promotional calendar for a beer brewing giant

• The aim of the project was to come up with an algorithm that would produce a Promotions Calendar for the Latin America arm of one of the world's biggest Beer Brewing Organizations
• Since the objective was to boost revenue/profit/market share which contain both linear & non linear metrics, Linear Optimization algorithm along with Genetic Algorithm were used to used to solve the problem
• A web application using Vanilla JavaScript and Django was built for seamless usage of the features