Hi, I'm Evan Brown.
A
Self-driven student passionate about finding new ways to observe data and predict outcomes.
About
I am a senior at Rice University pursuing bachelor's degrees in computer science and statistics, with a minor in data science. I have always loved finding hidden information in data, and my degree and internships have helped me develop the skills necessary to pursue this passion in my career. I believe my strength in both data science and computer science enable me to tackle complex data problems in a clean, efficient, and well-designed way. I am skilled in predictive model design/development, regression, bootstrapping, linear algebra, and experiment design.
- Languages: SQL, Python, R, Go, Javascript, C, HTML, CSS
- Databases: Databricks, SSMS, Azure SQL, Amazon Athena
- Data and Machine Learning Tools & Technologies: mlflow, TensorFlow, Spark, scikit-learn, XGBoost, Random Forests, Pandas
- Visualization Tools:Tableau, Retool, Streamlit, Plotly
Looking for an opportunity to develop new ways to understand data and find solutions to modern day problems in a responsible way.
Experience
- I built interactive user interfaces in Retool using SQL and JavaScript, including performance radars, optimized player searches, and team organization visualizations.
- These tools were used directly by stakeholders and clients to guide player and coach signing decisions.
- I started the internship with no experience in Retool. I taught myself as much as possible and learned how to use many new tools. I developed to the point where I served as the primary contact for all Retool issues. I authored comprehensive, company-specific documentation to streamline onboarding and troubleshooting, including common issues and fixes to company specific issues.
- I created historical, visual, and numerical analyses to assist in scouting and team success.
- I reported directly to the C-Suite, ensuring interfaces were clear and actionable, and aligned with strategic objectives.
- I had the opportunity to build a machine learning model from scratch to predict renewal likelihood for season ticket members, enabling the service team to prioritize high-impact accounts throughout the season. I built the model from the ground up, from design to construction to optimization.
- I engineered the model's features in an Azure SQL database to support model training.
- I developed and evaluated five models in Databricks, optimizing log-loss for accurate account-level renewal probabilities, helping the service team target key accounts with greater precision.
- This led to identifying an XGBoost model, optimized via Bayesian hyperparameter tuning, as the best-performing approach, achieving 87% average precision and 0.41 log-loss. This was a a 10% increase in average precision and a 0.15 reduction in log-loss over a simple baseline logistic regression model.
- I automated weekly prediction refreshes with a Databricks pipeline and stored results in Azure SQL for analyst use and Tableau integration. I used this to design and present a Tableau dashboard for the Service and BI teams, supporting targeted outreach and revenue forecasting.
- I ensured that the entire project was developed with in an organized, efficient, and well documented manner to assist in handover. I kept clean and document code for every stage of development, and included it in a clear and consise handover document.
- Designed and launched a website to provide free access to high-quality soccer coaching materials for anyone.
- Produced, edited, and authored 50+ in-depth videos and articles, reaching over 3,000 users in 50+ countries.
- Planned and led in-person trainings with 5-10 players at a time.
Projects
An analysis on player decision making using xG (expected goals) of certain passes.
Optimization of EURO teams' expected points based on unique rules of a friendly competition.
Predicting season ticket renewal probabilities of current members.
- Built entire pipeline, from data wrangling to an automated Databricks pipeline to a Tableau report for stakeholders.
- Focused on clarity, accuracy, and avoiding false positives. Clarity allowed sales team to understand which features caused members to not renew. Accuracy allowed for better reveneue projections. Avoiding false positives helped avoid not focusing on at-risk members.
- Compared multiple predictive models, tested out different features, and selected final feature list based on SHAP importance.
- Project was under an NDA. Cannot provide source code.
- Tools: SQL, Python, Databricks, Tableau
A model focused on selecting patent-reference similarity.
- Researched transformers, embeddings, and other NLP topics I previously did not know.
- Tested out many embedding strategies and distance learning models.
- Worked directly with stakeholders to decrease the number of references submitted without manual labeling. Has the potential to save thousands of dollars and hundreds of hours.
- Project was under an NDA. Cannot provide source code.
- Tools: Python, HTML, FastAPI, Jinja
A group project focused on project design, concurrency, and reliability.
- Built a custom database and a RESTful API in the backend using Go. Tested with Swagger.
- Focused on building a concurrent skip-list with extensive testing.
- Built a working messaging web app that worked across devices concurrently. Focused on a high-quality MVC design. Tested extensively with cypress.
- Project was part of a class and under Rice's Honor Code the soure code cannot be published.
- Tools: Go, HTML/CSS, TypeScript, Cypress
Education
Houston, Texas
Degree: Bachelor of Arts in Computer Science
Degree: Bachelor of Arts in Statistics
Minor: Data Science
GPA: 3.71/4.0
- Data Science – Tools & Models: Data analysis with SQL, optimization- and probabilistic-based learning methods, big data analytics and programming, AWS and storage infrastructure experience.
- Concurrent Program Design: Developed a fully functional messaging web app using HTML/CSS and JavaScript, along with a RESTful database written in Go to store the messages.
- Algorithmic Thinking: Performed mathematical reasoning about algorithmic problems, focusing on the design, efficiency, and accuracy of algorithms; explored search trees, hash tables, and other useful data structures.
- Rice Program for Accessible Working Service-Dogs: Event Planner and Board member. Helped organize meetings and plan the early steps of the program.
- Sport Representative for residential college: promoted and organized intramurals for residential college, nearly doubling participation.
Relevant Courseworks:
Leadership and Service:
Evergreen, Colorado
Salutatorian
Accolades:
- 4.7/4.0 GPA
- 36 ACT; 1530 SAT
- Four-time gold honor role: achieved a 4.25 or higher GPA on the 4.0 scale every year.
- Three-time varsity soccer letterman: three-year varsity starter, team captain as senior
Leadership and Service: National Honor Society Technology Coordinator/Officer. Helped run a 100-member chapter.
Varsity Soccer: Captained a 4a program to a top 3 league finish.
Relevant Coursework: 10 AP Courses including statistics and computer science principles