Total 650 words used.
Dr. Athanasios Lolos is a Senior Data Scientist at Navy Federal Credit Union (NFCU), the largest credit union in the world, serving over 14 million customers and employing over 30,000 people. With a PhD in Operations Research and an MBA from Georgia Institute of Technology, Dr. Lolos is an active researcher in the Operations Research field, focusing on quantile estimation. He also reviews papers for reputable journals like ACM Transactions on Modeling and Computer Simulation (TOMACS).
From 2023 to 2024, Dr. Lolos achieved significant milestones with impactful contributions both at NFCU and in the broader scientific community.
Dr. Lolos developed NFCU’s first-ever Financial Health Model, which assesses customer financial health and produces two metrics: a financial health score and a confidence score for data quality. These metrics enable personalized solutions for customers to reach their financial goals. The model has been used in several applications such as creating new customer segments and improving other models (e.g., boosting a predictive model's accuracy by 20%). The Financial Health Model is being considered as a key enterprise performance metric for NFCU’s strategic goals.
He created a new decision tree model that predicts overdraft transactions. This model has significantly improved accuracy (65%), compared to its predecessor (10%) and it is part of NFCU’s effort to provide personalized customer experience, preventing overdraft events and increasing customer satisfaction.
He proposed a new Certificate product with a bonus feature to discourage early withdrawals, thus increasing customer retention. Additionally, he developed an application with graphical user interface (GUI) for creating competitive certificate products, streamlining the existing process.
Dr. Lolos laid the groundwork for optimal weight selection for combining forecasts from different models, cutting forecasting errors (e.g., 12% to 5%). He also pioneered model monitoring advancements, including a new method for setting adjustable performance thresholds to account for issues related to the Mean Absolute Percent Error (MAPE) metric, new MAPE approaches for isolating errors of input variables and modeling methodology, and automated model monitoring report generation, saving over 500 hours in the first months of its use.
These initiatives have significantly improved NFCU’s ability to serve customers more effectively by using data-driven insights to create personalized experiences, ensuring a more efficient outreach overall.
However, Dr. Lolos' contributions extend beyond NFCU into the quantile estimation research field.
His SQSTS procedure for sequential quantile estimation sets a new industry standard, providing a highly efficient, automated tool for estimating confidence intervals (CIs) for quantiles with applications in industries like manufacturing and call center operations. SQSTS outperforms its competitors by requiring significantly smaller sample sizes. An example of SQSTS’ notable performance is presented in Table 3 (SQSTS paper) where SQSTS met the required performance using 378,815 data points while its competitor required 9,809,640. This efficiency is a huge benefit in situations where data acquisition is costly or time-consuming.
Additionally, he introduced FQUEST, the first fully automated fixed-sample-size procedure for computing CIs for quantiles. FQUEST performs exceptionally well even with small sample sizes, delivering CIs with coverage probabilities close to the specified nominal level. This innovation opens up new possibilities for applications like performance evaluation of manufacturing systems and risk analysis of financial portfolios.
Through his technical expertise, Dr. Lolos has not only contributed to NFCU’s customer-centric services but has also made profound impacts in quantile estimation methodologies, advancing the industry and expanding the scope of real-world applications. His work reflects his dedication to improving both business and academic domains, combining cutting-edge research with practical solutions.
Dr. Lolos’ contributions from his work at NFCU, and his groundbreaking research in quantile estimation make him an exceptional candidate for the Technical Professional of the Year award. His innovations have provided NFCU with tools to better assist customers in reaching financial wellness and have introduced state-of-the-art solutions in quantile estimation for business applications. With both his industry impact and academic recognition, Dr. Lolos stands as a clear leader in his field.
Nomination Title: Dr. Athanasios Lolos
