Aug 2022 – May 2023 · 9 mos

• Explored commercial opportunities around a project I help maintain nbdev: https://nbdev.fast.ai/

Sep 2019 – Aug 2022 · 2 yrs 11 mos

• Working on various open-source projects. My focus is on infrastructure, developer tools, DevOps, and software engineering.
• Some products I've been involved with:
• nbdev: A notebook-enabled software development environment - github.com/fastai/nbdev
• fastcore: Extensions to the python programming language - github.com/fastai/fastcore
• ghapi: A python client for the GitHub API w/ support for GitHub Actions - github.com/fastai/ghapi
• fastpages: A blogging platform with support for Jupyter notebooks - github.com/fastai/fastpages
• I've also created and help maintain many of the CI, integration tests, and build tools for the fastai ecosystem. Some examples include Conda build tooling https://github.com/fastai/fastconda, Docker image maintenance https://github.com/fastai/docker-containers, and many others.

Jan 2022 – Aug 2022 · 7 mos

• Building ML Tooling with Metaflow.
• Left to explore commercial opportunities with fastai & nbdev

Oct 2017 – Jan 2022 · 4 yrs 3 mos · San Francisco Bay Area

• Created systems and conducted research that enabled representation learning of code, issues, repos, and users.
• Led launch of CodeSearchNet (www.github.com/github/CodeSearchNet)
• Machine Learning Ops & GitHub Actions integrations: https://www.mlops-github.com
• Open sourced several examples of how to build data products and use deep learning with public GitHub data. These examples can be viewed at: http://hamel.dev, and include
• Summarizing GitHub Issues using sequence to sequence models
• Natural Language Semantic Code Search
• How to create machine learning enabled GitHub Apps to automate the developer workflow
• How to achieve Machine Learning Ops with GitHub Actions and Kubeflow

Jan 2016 – Jan 2017 · 1 yr · San Francisco Bay Area

• Led the creation of a company-wide machine learning infrastructure roadmap.
• Optimized marketing spend of $500M annually using machine learning.
• Built customer lifetime value models that predict spending patterns for every user at an individual level. This is used across the entire company in the following areas: marketing, customer service, risk and finance.
• Created image classification models to recognize objects in listing photos and for image re-ordering.
• Built end-to-end pipelines for data products using Airflow.
• Created reusable frameworks and infrastructure that enable other data scientists to iterate faster.
• Used a large suite of tools for modeling: Spark, Python, Keras / TensorFlow, DataRobot, H2o, Vowpal Wabbit.
• http://nerds.airbnb.com/data/

Jan 2015 – Jan 2016 · 1 yr · Greater Boston

• Data scientist and product manager for a machine-learning software company.
• Deployed a wide variety of modern algorithms such as xgboost, random forest, gradient boosted trees, support vector machines, elastic net etc. on operational systems.
• Regularly utilized python, spark and H2O for data wrangling, model building and visualizations.
• Built data science workflows in order to ensure that data was aggregated, collected and cleaned in ways that maximized accuracy and transparency.
• Contributed significantly to the product design and user experience for the data ingest and python API components of the product.
• Benchmarked and studied competing and complimentary machine learning software for accuracy, speed and usability. (H20, Graphlab/Turi, Amazon, Azure, etc.)
• Contribute to product strategy and roadmap.
• Presently an advisor on product strategy and data science.

Jan 2011 – Jan 2015 · 4 yrs · Greater Boston

• Led data science efforts on consulting engagements for a wide variety of industries. Some concrete outcomes:
• Founding member of the advanced analytics consulting practice, grew a team of 5 to 50 machine learning and analytics practitioners.
• Built data-driven tools that enabled a large restaurant chain reduce food-waste by 15% through demand forecasting.
• Created models and interactive data visualizations for a large casino to increase the detection of fraud by 50%.
• Performed market basket analysis and built recommendation systems that increased conversion rate by 5% for a large clothing retailer.
• Launched professional development curriculum for junior data scientists that built skills for data wrangling, predictive modeling and visualizations using Python and the command line.

Jan 2004 – Jan 2008 · 4 yrs · Dallas-Fort Worth Metroplex

• Specialized in CRM data analysis for fortune 500 telecommunications companies. Lead teams of 3-5 people on multiple engagements to implement data analytics solutions for clients in the telecom industry. Saved clients over $85M dollars in the areas of supply chain, call center optimization and churn reduction.
• Some examples include:
• Provided early detection of product defects and network failures through use of call center analytics, saving over $10M in contact center overhead.
• Built predictive models to provide an early warning system for customer churn, allowing client to intervene and reduce churn rate by 7%.
• Developed scorecards, metrics, and dashboards that enabled clients to increase customer satisfaction while simultaneously reducing costs by changing employee and partner incentive structures.

May 2003 – Apr 2004 · 11 mos · Greater Seattle Area

• Developed and implemented models to evaluate credit risk on loan portfolios.