My professional experiences below showcase my ability to bring ideas to life, whether it’s creating MVPs from scratch or launching brand-new projects within established companies. From initial concept to building and leading teams around these initiatives, I have consistently delivered results that drive impact. While these examples highlight my capabilities in professional settings, my personal projects offer an even deeper glimpse into my creativity and passions.
Exametric • Software Engineer • 2003 – 2006
Exametric was my first job in the U.S., marking the beginning of an exciting chapter in both my professional and personal life. After working for a bank in Istanbul for three years, I came to San Diego to pursue an MBA. Unfortunately, financial challenges forced me to leave the program early. Through a friend's recommendation, I secured a position at Exametric on an H1B visa. The company was founded by Ali Kiran, who was familiar with my prestigious alma mater in Turkey, which helped establish a connection.
My time at Exametric was transformative. Not only did I hone my skills in optimization, forecasting, and algorithm development, but I also found personal milestones — I met my wife, got married, and eventually became a U.S. citizen. Exametric was a forward-thinking company, implementing advanced concepts ahead of its time, and it provided me with invaluable experience that laid the foundation for my career in the U.S. The company was later acquired by Verint.
Exametric provided staffing optimization solutions for banks, including Bank of America, MFCU, and WestPac. The platform utilized advanced forecasting to predict transaction volumes by type and time, enabling banks to efficiently schedule staff.
The forecasting process primarily involved polynomial regression. Seasonal and holiday adjustments were applied on top of that. Queueing theory was then employed to determine the optimal number of staff required based on transaction predictions.
Optimized Regression: While regression is typically solved using gradient descent, I identified a faster approach by deriving the optimal solution mathematically through polynomial equation solving. I implemented this solution as a stored procedure in SQL Server without loops, significantly increasing performance by eliminating the bottleneck of slow loop execution.
Analyzed Forecast Algorithms: Conducted extensive research and development by analyzing and testing 20+ forecast algorithms. This work enhanced the robustness of the forecasting system and expanded its capabilities.
Fourier Forecasting Algorithm: Developed an alternative forecasting algorithm using Fast Fourier Transforms (FFT). After benchmarking it against the production model, I found its accuracy to be on par, providing a viable backup solution.
Hyperparameter Optimization: Created a mechanism to optimize hyperparameters for the forecasting methods, which resulted in a 40% improvement in forecast accuracy. This innovation played a key role in a forecasting contest, ultimately securing a $2 million contract with WestPac.
OLAP and Data Mining: Introduced an OLAP data warehouse component, which significantly sped up Bank of America’s forecasting process. The runtime for their forecasts was reduced from 6.5 hours to just 97 minutes, enabling more timely decision-making. This database also laid the groundwork for future expansions, such as new algorithms, advanced reports, and data mining operations.
Optimized Queries and Import: Enhanced SQL queries and optimized Data Transformation Services (DTS) packages. Results included a 5x increase in data import speed and over 100% improvement in forecast performance.
Technologies: VB6, C#, IIS, MSSQL, Analysis Server, OLAP, DTS, SSIS, MDX.
