I consulted independently in data science and machine learning, interfacing with clients both directly and via consulting platforms. I managed all aspects of my business - invoicing, taxes, etc.

I partnered with Spring Labs to develop the Income Stability Indicators (ISI), a novel data product which is poised to make the biggest change to financial underwriting decisions since credit scores. Working as a fully-embedded consultant, I collaborated with a small team of engineers and product managers to build the ISI data pipeline and models. I wrote production code in both R and Python, and I developed models using xgboost. In addition to software development, I also coordinated internal projects and presented directly to Spring Labs's clients which include major financial institutions.

Ibotta offers the world-leading platform for earning cash back on everyday purchases. As an engineer on the dedicated machine learning team, I used Python and Spark (running on top of AWS and Qubole) to build high-quality data products. Our team followed Agile development practices, writing production code with robust QA.I built an end-to-end, automated framework for continuous validation of online recommender systems which uses PySpark for data wrangling, standard Python libraries (pandas, numpy, etc.) for data manipulation, Airflow for automation, and DataDog for data visualization and monitoring. I wrote highly extensible code using best python and data engineering practices.I also spent time analyzing recommendation engines to understand how decisions made during the modeling process (choice of target variable, error metrics used in validation) corresponded with business KPIs such as retention and customer satisfaction.

Eaze is the groundbreaking cannabis technology company which facilitates a three-way delivery marketplace between consumers, dispensaries, and brands/suppliers.I used machine learning to improve Eaze's core operational algorithms: accurately predicting delivery times, optimally matching orders to drivers, and ensuring an appropriate level of driver supply via automated shift planning.I provided analytical leadership and support to the rest of the organization to improve the company’s collective understanding of customer satisfaction, customer retention, lifetime value, etc.Finally, I built data tools for the growth marketing team to enable rapid and robust experimentation and analysis. Our growth experiments ranged the gamut from email marketing to live operations, and we used quick opportunity sizing to prioritize effectively.

HomeAdvisor is the world's largest digital marketplace for home services, offering matchmaking and other tools for homeowners and contractors.I built, led, and managed HomeAdvisor's Applied Data Science team, responsible for day-to-day operations including hiring as well as long-term prioritization and planning. I communicated frequently with executives, department heads, my own reports, and other firms in order to ensure our team worked effectively toward HomeAdvisor's overarching goals. Individually I provided statistical and mathematical expertise to groups throughout the enterprise.My team's primary focus was to support and drive initiatives in Product, Marketing, and Operations. On the Product side, we worked towards optimizing the site's matching algorithms using machine learning techniques, large-scale computing, and mathematical analysis. We worked in conjunction with HomeAdvisor's software development teams to deploy models into the production system and ensure models were working accurately and efficiently.Our team was responsible for the analysis and optimization of HomeAdvisor's various marketing channels including television/media, search engine marketing, and email marketing. We designed and analyzed several tests a year, and we developed an in-house test design and analysis platform on par with enterprise software that typically costs hundreds of thousands of dollars per year.For Operations, we began building reports and systems to support both our sales force and our customer service force, each of which was comprised of hundreds of employees. Our goal was to determine the optimal allocation of resources such that the network operated as productively as possible towards net revenue, profit, and customer satisfaction.Note: For the first five months of this position (December 2016 - May 2017), my title was Associate Director, Business Intelligence.

I managed the algorithms on HomeAdvisor's website which determine the viability and utility of various homeowner-to-contractor matches. I made use of machine learning algorithms such as random forest, GBM, support vector machines, logistic regression, and others in order to refine the matching experience. My improvements led to substantial increases in customer contact rate and net revenue.I facilitated the implementation of machine learning and artificial intelligence into HomeAdvisor's production pipeline via AWS servers running R, Python, Java, and Hadoop. I explored the use of H2O and other massively parallel tools in order to optimize HomeAdvisor's existing workflow and offer new possibilities moving forward.

I used SQL, R, and Excel to generate reports which were used to inform decision making across the business, particularly with respect to corporate strategy.I built clustering models for customer segmentation, forecasting models based on time series decomposition, and I analyzed the results of A/B and market-based tests with uneven groups.Note: For the first three months of this position (September 2013 - November 2013), my title was Data Analyst Intern.

I researched the application of probabilistic graphical models to election prediction on a grant from Rice University.

I graded problem sets for a course in Computational & Applied Mathematics.

I completed a large-scale mortality study using Microsoft Access and Excel, making various improvements to the methodology in the process.I worked with other interns to document our learning experiences into a booklet that could be used by other employees in the future.