Andrew Clarke

About Andrew Clarke

I’m an experienced data scientist with a 10 year background of delivering a diverse set of data science products that drive strategic impact in a variety of industrial settings. These include production-ready machine learning models including LLM implementations, org-level measurement frameworks, and cold-start experimentation at scale for enterprise level organizations. I use both R and Python for statistical analyses and model-building via Apache Spark in Databricks, have extensive database querying experience via SQL from a variety of database types including Snowflake and Redshift, and am comfortable deploying both machine learning models and other ETL processes both batch and in real time via the AWS suite of tools (S3, ECR, Batch, Lambda) as well as Apache Airflow. I use Tableau, RShiny and Python Dash for data display and reporting, and the full suite of Atlassian tools for progress tracking and documentation. For experimentation, I have significant experience with Statsig, and have run the deployment of this platform for Atlassian\'s customer service organization.In addition to driving strategic impact as an individual contributor, I also have a proven track record of mentorship, project leadership, and personnel management in industry and academia. When I\'m not working, I\'m rowing on Lake Samamish with the Samamish Rowing Association.

Experience

1. Principal Data Scientist

   Atlassian

   Aug 2022 — Present · Seattle, WA, US

   Built the experimentation program for the Customer Support Services (CSS) division from the ground up, including a comprehensive measurement framework for key OKRs, ETL transformations to report on the same, and full integration with the third party feature gating and experimentation platform Statsig. Also designed and deployed tools for experiment planning including a Monte Carlo simulation-based power analysis module and a time-series forecasting module for predicting experiment enrollments. Quadrupled feature testing and deployment in FY2024 compared to FY2023, and realized effort savings of 20%, totaling $25M in headcount savings for FY2025.Pioneered quasi-experimentation, attribution modeling, and causal driver analysis within the Customer Support Services (CSS) division to quantify the effects of customer support SOP changes on cost-to-service, demonstrating a cumulative $55M reduction in support costs in FY2023.Launched CSS\'s first customer-facing AI agent, and developed Atlassian\'s first AI measurement framework. This unified approach to evaluating AI agent performance and efficacy has been adopted throughout CSS to streamline AI agent optimization.

Education

• University of Alabama at Birmingham

  Doctor of Philosophy - PhD, Bioinformatics, and Computational Biology

  2013 — 2019

• University of Puget Sound

  Bachelor's of Science, Chemistry

  2009 — 2012