● Led supply chain network design and optimization project to create analytical solutions and provide recommendations to leadership with detailed financial impact summary, uncovered over $3M saving opportunities across network, prototyped a web application using Python Flask for the optimization model built in Pyomo. ● Applied Self-organizing mapping model to cluster products based on historical demand pattern and recommended similar inventory stocking strategy as well as provided forecasting model at product cluster level.● Led hackathon project with a team of 10 people across IT and Analytics teams to create a multivariate time series forecast model with external datasets such as COVID confirmed cases to predict demand for certain drug categories and customer geography group with MRSE about 10%. ● Applied linear regression model to estimate labor hours for 20 distribution warehouses using several outbound throughput volume metrics with R-square score over 90%. ● Custom built a multi-echelon inventory replenishment model in Python to simulate different stocking policy and its impact on inventory and labor cost in distribution warehouses.

● Worked on the new product onboarding automation project with a team of 10 people on Google cloud platform. Created a rule-based method with OCR API to extract text field from tabular documents with 75% accuracy, and a logistics regression model using Scikit-learn with tf-idf vectorizer transformation on texts from pdf file with 95% accuracy. ● Created an optimization model in Python with CPLEX to solve the multi-echelon inventory replenishment schedule problem. Uncovered significant saving opportunities considering the trade-off between inventory, transportation, and labor cost. ● Applied clustering model to identify optimal ranges of inbound volume by product, custom-built a discrete event simulation model in Python to simulate warehouse receiving, put-away and picking activities at product level to estimate the quantity and right sizes of storage locations. The recommendations increased storage utilization by 30% and freed up about 6M cubic inches in space in one of the pilot warehouses. ● Applied random forest classification model to predict suppliers’ inbound delivery lead time with accuracy over 75%, projected to increase inbound labor planning accuracy significantly due to improved volume estimation.

● Participated in the pre-sales activities of flexis core strategic planning products for global automotive OEM’s, formulate the end-to-end manufacturing process flow using internal production variables. Mapping from customer’s business requirements to available and potential product functionalities, creating the adaptation plan of moving from standard product to customized solution. ● Worked with a team of 6 software engineers on product feature development, existed modules enhancement written in Java, JavaScript, Python, HTML and CSS.● Worked with Research and Development team on production sequencing module enhancement with constraint programming in ECLiPSe.

● Developed algorithm to solve Markov decision processes with uncertain transition probability and reward with optimality guarantee, applied the algorithm in Cyber vulnerability maintenance planning with data collected from 512 host computers across three labs of The Ohio State University, and recommended optimal policies to potentially prevent 40% of the cost with maintaining cyber security. ● Developed algorithm to solve real-time job shop scheduling problems with uncertain job arrival and duration time, validated the quality of solution with numerical examples and math proofs. ● Researched and published a literature review on job-shop scheduling algorithms considering material handling in between jobs. Implemented a metaheuristics algorithm in real-life project to help manufacturing plant create daily production schedule automatically with saving 80% senior planning manager’s time from manual planning.