Andrew Tilley, Data Science Developer in Denver, CO, United States
Andrew Tilley

Data Science Developer in Denver, CO, United States

Member since October 24, 2019
Andrew has over seven years of experience deploying predictive algorithms and end-to-end data solutions, increasing his companies' net revenue by tens of millions of dollars. His technical acumen, combined with deep communication skills to both technical and non-technical colleagues, ensures he has a broad impact across teams who work with him. Whether the end result is a data product or a detailed analysis, Andrew relishes the opportunity to solve business problems with machine learning.
Andrew is now available for hire

Portfolio

Experience

Location

Denver, CO, United States

Availability

Part-time

Preferred Environment

Python, R, SQL, Spark

The most amazing...

...product I co-invented enabled millions of connections to happen in minutes rather than days.

Employment

  • Machine Learning Engineer

    2019 - 2020
    Ibotta, Inc.
    • Built an end-to-end framework for continuous validation of online recommender systems using Python, Spark, and AirFlow. Utilized best engineering practices leading to an extensible, highly-interpretable program.
    • Contributed utility functions for common ETL tasks relating to online metric tracking via Python and DataDog.
    • Analyzed recommender systems to understand the impacts of modeling decisions on downstream business metrics such as retention, re-orienting the team’s process, and improving future strategy.
    Technologies: Python, Spark, SQL
  • Senior Growth Scientist

    2017 - 2019
    Eaze, Inc.
    • Owned delivery algorithms and invented key algorithm concepts, which improved ETA accuracy by 50%. Led relationships with product and engineering teams to plan and implement algorithm changes.
    • Built a self-service experiment analysis and visualization tool empowering the marketing team to run email experiments in seconds rather than days.
    • Automated driver shift planning models to reduce driver costs while maintaining delivery times.
    • Advised and guided other data scientists on mathematical models and relationship building with other departments, leading to improvements in team output and cohesion.
    Technologies: R, SQL
  • Associate Director, Data Science

    2013 - 2017
    HomeAdvisor, Inc.
    • Earned multiple promotions from data analyst to technical product manager and finally associate director in December, 2016.
    • Founded, recruited, led, and managed a six-person data science team which made key optimizations to the matching engine and increased the ROI of marketing spend using regression-based analysis.
    • Owned matching algorithms and invented new ones utilizing machine learning (regression and tree- based methods) that improved net revenue by 5% and customer contact rate by at least 25%.
    • Developed an in-house test design and analysis platform on par with enterprise software that typically costs hundreds of thousands of dollars per year.
    • Managed cross-functional projects with teams from technology, marketing, product, and finance to ensure predictive algorithms met the needs of all stakeholders.
    Technologies: R, SQL, Excel

Experience

  • Instant Connect Model for HomeAdvisor (Development)
    https://pro.homeadvisor.com/articles/videos/instant-connect

    "Instant Connect" is an experience on HomeAdvisor's website, which enables a homeowner to get connected over the phone to a contractor as they are submitting their service request. I was the lead data scientist on the project and a co-inventor on the patent. My role was to build the models which would determine (1) when it made sense to present the "instant connect" option to the homeowner, and (2) which contractors should be called and in what order. I used a combination of statistical regression and applied mathematics to ultimately improve the contact rate by over 25%.

  • Recommender Performance Tracker for Ibotta (Development)

    I built an end-to-end, automated data pipeline, and visualization for the tracking of Ibotta's online recommender systems. The program (written in Python and Spark) was completely decoupled from the model-building programs, which meant it provided an unbiased view of real-world model performance. By inventing a new layer of abstraction to compare recommenders to each other, I made the program highly-extensible, and more recommenders could be added with little overhead. I set up a dashboard in DataDog, which was used by machine learning engineers and analytics leaders to assess recommender performance.

  • Data Science Assessment for Eaze (Development)
    https://github.com/tilleyand/data-science-homework

    I wrote an original assessment of five questions spanning applied mathematics, statistics, and computer programming for prospective data science candidates of Eaze, Inc. The 3-hour assessment was used to filter dozens of candidates, ultimately leading to two highly-successful new team members joining Eaze's analytics team.

Skills

  • Languages

    R, SQL, Python
  • Libraries/APIs

    Ggplot2, PySpark, Scikit-learn
  • Paradigms

    Data Science, Agile Software Development
  • Platforms

    RStudio
  • Other

    Data Analysis, Regression, Optimization, Applied Mathematics, Machine Learning, Statistics, Combinatorial Optimization, AWS
  • Frameworks

    Spark
  • Tools

    Apache Airflow, Git, Microsoft Excel, LaTeX, MATLAB
  • Storage

    AWS S3, Redshift

Education

  • Bachelor's degree in Computational and Applied Mathematics, Statistics
    2010 - 2013
    Rice University - Houston, Texas, USA

Certifications

  • Building a Data Science Team
    OCTOBER 2016 - PRESENT
    Coursera
  • Hadoop Platform and Application Framework
    OCTOBER 2016 - PRESENT
    Coursera

To view more profiles

Join Toptal
Share it with others