Aaron Trefler Email & Phone Number

Los Angeles, Ca, Us

Los Angeles, Ca, Us

San Diego, Ca, Us

San Diego, Ca, Us

Des Moines, Ia, Us

• Created an end-to-end content recommendation system for MyLife, a mindfulness app, to drive free users towards paid subscriptions using SQL for data extraction; Python for data cleaning, analysis, and visualization; Scikit-learn and XGBoost for supervised machine learning; and SageMaker for real-time model deployment• Improved Bizrate’s magazine subscription recommendation system (50K visits per day) by developing a module to allow recommendations to be based on predicted lifetime value; creating a performance monitoring dashboard using automated Jupyter notebook executions and QuickSight; and re-factoring the ML pipeline codebase using EMR and Spark• Developed a solution for targeted magazine-offer post-cards by using Experian demographic data to create ideal magazine personas; dimensionality reduction (PCA) and clustering (K-Means) to map targeted individuals to their most relevant magazine post-card cluster; and t-SNE to show the validity of the approach to business stakeholders• Took initiative to always present completed projects to stakeholders and teammates in order to communicate insights, drive alignment, and allow for the synthesis of new ideas

Des Moines, Ia, Us

San Diego, Ca, Us

• Created Classy’s current full-cycle machine learning system (performance +20% over naive baseline model) to predict the success of non-profits in their first year using Classy’s online fundraising platform• Implemented predictive modeling system codebases (churn prediction, lead scoring) using SQL for data extraction; Python for data cleaning, visualization, and reporting; and Scikit-learn (and XGBoost) for model training and hyper-parameter tuning• Performed feature engineering by leveraging a wide range of business data sources (Classy Product, Salesforce, IRS Tax Form-990, social media statistics, Datanyze, Alexa Website Traffic) to increase the performance of supervised machine learning models• Directed weekly project management meetings, attended by 3 senior teammates, in order to ensure project deliverables remained on schedule, roadblocks were overcome, and insights were shared• Presented at Classy’s All Hands meeting (200+ attendees) on the impact of default donation settings on charitable gift sizes by creating a technical report and then synthesizing its key insights into a presentation suitable for all Classy employees (video below)• Implemented ETL processes using REST APIs (Charity Navigator, YouTube) in order to grow Classy’s AWS Redshift Data Warehouse• Advanced software development skills by regularly performing and receiving code reviews, practicing Agile, using JIRA to manage sprints, and leveraging Git and Bitbucket to version control all source code written

Pasadena, Ca, Us

• Conceived, designed, and conducted a data-driven investigation into the relationship between land-water-storage and global flood likelihood and magnitude in order to further NASA’s overarching goal of using geo-satellite data for active global flood monitoring• Merged, analyzed, and statistically modeled data from several disparate sources (NASA geo-satellites, The Global Flood Archive, and NOAA's precipitation model) using Python and Matlab in order to investigate the flood forecasting ability of NASA’s GRACE satellite• Showcased findings by completing a detailed project report and giving a presentation to several full-time doctoral level JPL staff

Bethesda, Md, Us

• Spearheaded an investigation, involving 2 NIH labs and 5 Ph.D.’s, to elucidate the confounding relationship between time-of- day and structural MRI brain metrics in order to improve the quality of future structural brain plasticity research• Self-started research resulted in a first-author publication (95th percentile on Altmetric in terms of attention received, from 6+ million tracked publications); 2 co-author publications (2018, 2019), academic conference presentations, and ongoing lab research• Automated data processing workflows on the NIH super-computing cluster by linking multiple neuroimaging software suites and parallelizing data processing workflows using Unix-Shell scripting• Performed data analysis and visualization using Matlab; linear mixed effects modeling using R; and data organization using Excel

• Performed research on developing automated classification tools to distinguish patients with epilepsy from those with psychogenic seizures by working with structural MRI and time-series EEG brain data using Matlab, Excel, and Weka• Research resulted in two co-author publications, as well as two first-author manuscripts published online by the UCLA Department of Psychology

Culver City, Ca, Us

• Provided individual and group-tutoring services for fellow college students at the college library's Human Learning Resource Center

Washington, Dc, Us

• Completed official government forms for a variety of unique housing situations• Efficiency on performance evaluation metric (i.e., completed forms per hour) led to selection for an enumeration team assigned to problematic housing cases

Master Of Science (Ms), Computer Science - Specialization In Artificial Intelligence

Postbaccalaureate Coursework, Computer Science

Bachelor Of Science (Bs), Cognitive Science - Specialization In Computing

Undergraduate Coursework, Math, Science, And Computer Science

Undergraduate Coursework, Computer Science

Data Science