Nathan Midkiff Email & Phone Number

United States

• Demonstrated end-to-end ownership of 7 projects in one year, creating scoping and design documents, implementing the code, creating unit and integration tests, and collaborating with subject matter experts, product managers, and UX designers.• Conducted A/B tests for each project I implemented, creating an experiment plan, running the test on millions of users on our team's page, and performing a post-experiment analysis. This analysis involved looking for any bias in the results, analyzing the impact of the change on business metrics, and digging deep into the positive and negative impacts of the change, in order to create durable learnings for future experiments.• Created a new feature, with over 18 million uses per month, and shared my results with other teams across Amazon, leading to the feature being implemented across the site.• Designed and implemented a new feature, with over 1 million uses per month, by understanding customer behavior using the learnings of a previous analysis.• Participated in an oncall rotation, including handling over 10 high severity issues, during which I would dig into issues found on our site, including bugs, latency, and bot tra c, as well as being available 24/7 to handle high-severity issues.

Verona, Wisconsin, United States

• Completed 3 months of training in Python, C#, Typescript, and SQL which led to knowledge in software development practices.• Worked in a team of 12 to design and build predictive models for an Inpatient hospital setting for Epic customers.• Improved accuracy of these models by 17% by designing, writing, and testing SQL scripts which collected patient data for use in data models. • Created a data model pipeline in Python which cleaned and transformed data and trained a model to predict readmission among diabetes patients.• Designed and implemented retrospective validation for an ICU predictive model used in 29 hospitals around the US through collaboration with a cross-functional team of model and framework experts.

Los Angeles, California, United States

• Current member of UPE, an International honor society in the Computing and Information Disciplines, with members among the most outstanding undergraduate Computer Science students at UCLA• Tutored over 20 engineering students at UCLA in a wide range of subjects, including computer science, math, and physics• Refined skills in networking, web services, and databases through professional workshops in Microsoft Azure and cloud computing

Los Angeles, California, United States

• Conducted research investigating the spread of COVID-19 in the United States, providing accessible data and predictions for public viewing• Collected data on hospitalization and infection rates using Python, web scraping tools, web APIs, and data visualization tools, in order to keep a regularly updated database for use in understanding and developing models for these rates• Collaborated with a team of undergraduate and graduate students to form a machine learning model to predict future infection rates and improved SIR model• Visualized the results of the predictive model using including two maps to display the reported and true infection rates and a graph for understanding when the model believes peak cases will occur

West Lafayette, Indiana, United States

• Conducted research investigating the use of sparse neural networks as a method to reduce parameters in a neural network using Tensorflow and Pytorch• Programmed basic models such as multilayer perceptrons and convolutional neural networks to better understand Tensorflow and Pytorch• Developed a customizable neural network model using the Pytorch API to perform tests on sparse neural networks• Executed a pipeline to build multiple sparse neural network models and test and compare performance• Authored a comprehensive report describing findings in studying sparse and dense neural networks, discussing the problem,describing the procedure, and explaining the results of the work done• Connected with Purdue's community of professors and graduate students through attending weekly lectures and workshops coveringmultiple topics about neural networks