Mehmet Selim Çetin

• Engineered automated data workflows using KNIME, reducing manual processing time for repetitive tasks by approximately 40% (~10 hours/week). • Conducted error analysis on SAP master data using SQL, identifying data inconsistencies and improving overall data integrity metrics by an estimated 15%. • Collaborated with cross-functional teams to streamline data migration to a unified SAP platform, ensuring compliance with new data governance standards.

• Leveraged Python (Pandas, Scikit-learn) and statistical methods to analyze customer usage data, generating insights that increased product feature adoption by an estimated 20%. • Implemented machine learning algorithms to solve client-specific optimization problems, contributing to a 10% reduction in operational inefficiencies across key accounts. • Translated complex stakeholder requirements into technical data specifications, accelerating delivery cycles by approximately 2 weeks per sprint.

• Validated and back-tested predictive credit risk models, applying statistical tests that enhanced model accuracy (Gini coefficient improvement) by approximately 3%. • Developed comprehensive data validation frameworks across 5+ risk models, ensuring robust performance evaluation and regulatory compliance. • Automated model monitoring reports using Python and SQL, reducing monthly reporting lead time by approximately 30%.

• Built and evaluated machine learning models to optimize cash forecasting for bank branches, targeting reduction in idle cash holding costs. • Led a team of interns to develop an automated reporting tool using Jira data, cutting daily status meeting durations by 30%.

• Conducted statistical efficiency analysis of software testing processes, implementing data-driven recommendations that reduced bug recurrence rates by 10%. • Optimized the development workflow based on process mining, resulting in a smoother test design cycle

• Analyzed production data to identify process inefficiencies and propose optimization strategies. • Collaborated on improvement projects to increase line efficiency and reduce production costs.

Thesis Title: "History Aware Sampling for Robust Deep Learning: A Generalized Framework for Handling Noisy Data Across Classification and Regression Tasks" https://www.dropbox.com/scl/fi/hmgjo3tf1uy08717uzuzn/Thesis_final.pdf?rlkey=6rps2melji2z47ygaffx0r1ai&st=h4vqo7yx&dl=0

Cum Laude Full Scholarship