Mayukh Dewan Email & Phone Number

Waterloo, On, Ca

Toronto, Ontario, Canada

• Performed in-depth analysis on marketing data to identify trends and insights, leveraging statistical models to drive data-driven decision making and improve campaign performance.• Designed and launched synthetic control tests to test google hotel ads spend, resulting in a 6.5% growth in Net Revenue.• Automated the daily reporting process by designing and implementing a custom data reporting tool using Google Apps Script. This tool extracted data from Google Sheets and pushed daily reports to Slack channels via Slack webhooks.• Developed a Last Paid Touch Source model in BigQuery and built a looker dashboard adding Cost of Acquisition to the model, enabling the growth team to gain insights into the attribution and effectiveness of paid marketing efforts.

• Designed and developed a comprehensive Looker dashboard utilizing car GPS data to evaluate the level of service at intersections in variousUS cities. This dashboard provided valuable insights into traffic patterns and intersection performance.• Played a key role in enhancing the performance of existing Looker dashboards by optimizing derived queries and handled ad‑hoc requests toupdated looker dashboards.• Created a script to automate the execution of the signal plan optimization process and upload scenarios to S3, enabling efficient data analysisand visualization in Looker.• Ensured the accuracy and reliability of a traffic signal plan optimization tool by conducting thorough validation checks by comparing multiplescenarios against Synchro scenarios to validate the tool’s output.

•Built out the data and reporting infrastructure from the ground up using dbt,AdobeAnalytic and BigQuery to provide insights into the website experience and business KPIs.• Designed and developed a Looker dashboard for management to look at customer forecast KPIs and data.• Improved Website traffic forecasting algorithm to 87.9% accuracy by implementing and hyper parameter tuning a prophet time series model.

- Integrated into the client facing data science team - presented results to senior leadership and wrote an executive summary detailing value proposition and strategy to productionize model.- Lead on an IoT sensor imputation project to develop a time series solution in python. This model imputes missing data from IoT sensors and stored results in a SQL database.- Data visualization lead supporting the Product team: Created a product analytics Power BI dashboard using data from multiple SQL databases and created a schema for them.- Implemented a Web Scrapping solution to support the Media Sales team. This script pulls categories of 3000 different mall stores from google significantly removing manual effort on the sales team.

- Created an automated training pipeline which retrains the model on a periodic basis by adding new data to improve model performance and prevent concept drift.- Improved the model prediction speed by 10 times using more optimized data preprocessing algorithms and multi‑threading libraries.- Pulled 1.5GB of log data from ElasticSearch a type of NoSQL database, processed it for the model to predict on and pushed the predictions back to the database.- Used SOSL queries to query Salesforce for getting new data to update the training dataset, also implemented a hashmap to classify new logs as anomaly or not.- Implemented a design for the feedback loop which allowed the model to be retrained on the logs it predicted incorrectly.

Bachelor'S Degree, Statistics

Activities and Societies: Central Orientation Team 2022 @ WUSA Hackathon Research and Design Lead @ Schrodinger’s Hack Peer Success Coach.

High School Diploma