New York City Metropolitan Area
A certified Data Scientist, creative problem solver, striving to be critical, and pursuing high performance standards by persistently seeking for a new process in business operations.
• Increased purchase likelihood by 5% by implementing a transformer-based multi-label classification model on LG.com user behavior data, enabling precise audience targeting for return visits through comprehensive analysis of on-site interactions. • Summarized product reviews and extracted product-specific keywords using BERT, UMAP, HDBSCAN, and TF-IDF, enabling users to make informed purchase decisions by showcasing key insights on LG product pages. • Streamlined call center operations by transcribing audio files with NLP based model and building a custom question- answering model to identify caller intent, root causes, and solutions, improving resolution efficiency. • Deployed an automated customer segmentation pipeline processing 12M+ records across 1000+ features, enhancing customer targeting and driving personalized tailored marketing strategies to boost engagement and conversion rates. • Developed a Marketing Mix Modeling (MMM) tool to optimize marketing budget allocation, enabling data-driven decisions
• Optimized factory production processes by solving a convex optimization problem using cvxpy and Gurobi, balancing scheduling, raw material availability, and sales forecasts, leading to enhanced operational efficiency • Improved inventory forecasting by developing a scalable time-series clustering algorithm using Dynamic Time Warping to group products based on future sales forecasts.
Teaching Assistant for Machine Learning, Data Science, Deep Learning, and AI. - Designed convolutional neural network, computer vision model for classifying people’s age, gender, and race, in partnership with MotionFlow, a smart AIoT ad company
Automated demand forecasting procedure which incorporated probability of GI, weekly sales trend, open orders, and cancellation rates by region and SKU level, perform faster and accurate calculation measures. - Reduced daily reschedule rates under 2% YTD: by setting the procedure to prioritize orders by RDD, correct inventory consumption process by on-hand and in-transit inventory stock, and stabilized cross-dock shipment frequencies. - Enabled faster delivery: Analyzed root causes and adjusted to ideal delivery lead-time for PO creation to RDD per location. - Analyzed weekly KPI reports: for YoY account, product category level sell-out from BW BeX - Initiated controlling ideal inventory availability: to expand competitiveness in sales using national ATP.
Interpublic Group (IPG) System Maintenance: - Monitored and data flow and maintained errors process in ETL of OLAP cubes in SAP - Controlled the data flow in BW and maintained InfoObjects with Key Figures in needs.