Daniel Nathanael

Act as the sole data contributor in Payment Gateway business; from data warehousing, dashboarding, insight deliveries, experimentations, and ML modeling

Became a modern day detective who leveraged data to investigate, uncover, fight, and prevent financial fraud in Traveloka. • Decreased payment fraud rate from 0.04% to below 0.001% by implementing rule-based fraud detection system, with recall between 70%-80%, from doing Exploratory Data Analysis (EDA) and backtesting in BigQuery SQL and Python Notebook. • Prevented an 11% decrease in payment success rate from an immature policy by conducting end-to-end A/B experiments, including pre- analysis, experiment setup, and post-analysis. • Drove a 66% revenue growth from credit card installment by creating a pricing recommendation system using LGBM and Greedy Optimization. • Mitigated an annual loss of 68k USD from coupon abuse by designing an abuser detection algorithm that normalizes and cross-verifies user attributes like email, IP address, and contact information. • Saved 200k USD annually by replacing a vendor fraud scoring system with an in-house rule-based decision tree model. • Reduced fraud investigation time from a minimal 1 man day to half a day by creating a fraud monitoring and investigation dashboard in Looker Studio, integrated with data from ETL pipeline using DBT, Git, and Airflow.

Became the Subject Matter Expert (SME) in terms of data for logistics, collection points, and verification business units, from data ingestion, maintenance, tracking, and reporting. • Reduced daily reporting time from half a day to 2 hours by automating KPI performance tracking and reporting through a DBT ETL pipeline and dynamic dashboards in Metabase and Tableau. • Supported the expansion of 50 new Collection Points in Bandung Raya by visualizing strategic location coverage within specific radii using Metabase and BigQuery SQL. • Enabled the Chief Operating Officer’s Central Java expansion strategy by delivering a geographical heatmap dashboard of sales metrics in Tableau, highlighting regional opportunities. • Reduced discrepancies in COD (Cash on Delivery) payment reconciliation reports between finance and operations from 50% to near 0% by developing a Single Source of Truth (SSOT) table in BigQuery SQL, orchestrated with Apache Airflow and DBT, and automated reporting in Metabase., etc) correctly for stakeholders.

• Reduced the partnership business team's daily reporting time from a full day to just 3 hours by developing an automated dashboard in Looker Studio, leveraging DBT and BigQuery SQL. • Fully automated the cashback injection process, eliminating manual work by creating a tracking table in BigQuery, orchestrated with Apache Airflow and DBT, to manage cashback eligibility and amounts for customers. • Decreased payroll service cash outflow from 90% to 85% on payroll day by recommending targeted discounts and promotions, based on user behavior analysis conducted in BigQuery SQL and Google Sheets.

• Automated daily Liquidity Coverage Ratio (LCR) and Net Stable Funding Ratio (NSFR) reporting by leveraging MSSQL and Power Query in SQL. • Created monthly credit vintage analysis report using BigQuery SQL and Excel. • Fully automated credit risk daily reporting, from taking half a day to 2 hours, in a Looker Studio using BigQuery SQL. • Assisted POC backtesting for credit risk scoring done by vendors using Kolmogorov-Smirnov (KS) test in Python and Excel. • Assisted the creation of a logistic regression model to predict the Probability of Default and Estimated Credit Loss for credit loss provisioning and IFRS9 reporting in Python Notebook and Excel.

Laboratory assistant at elementary physics laboratory, which responsible of overviewing and guiding practician throughout lab activities, giving instructions, explaining theories, also reviewing and grading their reports.