• Business Requirements Analysis: Led and collaborated with cross functional team to comprehend the requirements for KPI metric data visualization• Data Collection: Collaborated with the data team to extract impression and survey data using Python and Jupyter/Bento Notebooks• Data Preprocessing: Utilized Python, logistic and KNN algorithm in Jupyter Notebooks for balancing test/control and bias correction• Key Metrics Calculation: Calculated metrics such as average, standard error, and lift (change in test/control responses across the campaign)• Experimental Hypothesis Testing: Conducted A/B tests with difference-in-differences (DiD) to evaluate the effectiveness of various campaign strategies• Observational Hypothesis Testing: Applied techniques such as propensity score matching to analyze the marketing execution by channel and channel mix• Statistical Testing: Evaluated in-campaign test/control differences and lift with a 90% confidence interval using Presto, SQL and Hive• Deployment and Monitoring: Developed dashboards for hypothesis testing (both observational and experimental), provided recommendation, and advised on media selection; engaged 70% of the target audience, enhancing KPIs and reducing costs mid-campaign, and later recommended a media switch, engaging 90% of the target audience, further improving KPIs

• Business Requirement Analysis: Led and collaborated with cross functional to understand business problem and establish KPIs• Data Collection: Collected historical transaction data from diverse sources using Hive/SQL, ensuring its relevance and adequacy for a CLV model• Model Selection: Selected a linear regression algorithm due to its suitability for the data and the project objectives• Data Preprocessing: Conducted profiling, cleaning, and standardization to enhance model performance. Partitioned data for training, validation, testing• Model Exploration/Optimization: Trained multiple regression models, cross-validated using resampling procedures for adjusted R-squared, tuned hyperparameters for optimization, and tested predictive accuracy and reliability using Python, Azure Machine Learning, VS Code and Jupyter Notebooks• Deployment and Monitoring: Implemented the CLV model in the production environment. Developed a Tableau dashboard for data visualization, enabling real-time monitoring and anomaly detection and improved customer retention by 10%

• Business Requirement Analysis: Partnered with the AML SME to understand business issues and defined metrics/KPIs• Data Preprocessing: Profiled and integrated account, customer, and transaction datasets for completeness, validity, and form a holistic view in SQL Server• Statistical Testing: Analyzed new transaction data using Student’s distribution to identify outliers, applying a 5% p-value for significance test using Python• Deployment and Monitoring: Created Tableau dashboard, conducted UAT reviews, and automated daily outlier reports to operations team

• Business Requirement Analysis: Partnered with SMEs to refine Critical Data Elements (CDE) necessary for 15 MLS feeds, covering 150,000 active listings worth about $80 billion, along with numerous rental listings• Data Collection: Collaborated with MLS to access their database via API, establish connections, and extract sample data for analysis.• Data Preprocessing: Analyzed geography-specific MLS feed data using Amazon Redshift and SQL to confirm Compass’s access to essential fields and evaluated data quality based on completeness, validity, and row count. If any issues were detected, coordinated with MLS to address them

• Business Requirement Analysis: Partnered with Data Stewards to identify CDE using business processes, thresholds, and data quality rules• Data Preprocessing: Standardized data using reference tables, applied quality rules, and increased summarization efficiency by 30% using SQL jobs• Key Metrics Calculation: Calculated data quality metrics such as completeness and validity ratios for the system

• Enhanced reports by automating data cleaning and SQL tuning, saving 5 hours weekly by eliminating manual spreadsheet processing

• Utilized SQL for data wrangling to enhance product quality, decreased data discrepancy from 9% to 0.07%, and increased product performance by 35%