Jul 2020 – Jan 2024 · 3 yrs 6 mos · Los Gatos, California, United States

• I lead the Model Development and Inference Infrastructure teams as part of the Machine Learning Platform at Netflix. We focus on making the process of taking ML algorithms from idea to production as efficient as possible, regardless of the size, scale, or shape of the problem. Our work powers mature large scale consumer facing ML use cases, experimental early stage 0-1 projects, and generative AI projects.
• My teams are responsible for easy end-to-end development and productization of ML pipelines, high performance internal training platforms and efficient, flexible inference frameworks. We use a combination of hardware, software, and ML engineering/optimization techniques to deliver a delightful experience to our users, and business value to Netflix.
• We partner closely with enterprise and open core partners Anyscale (Ray), NVIDIA, AWS, and Outerbounds (Metaflow) to maximize the leverage of external ML innovation for Netflix use cases.

Nov 2018 – Jul 2020 · 1 yr 8 mos · San Francisco Bay Area

• I managed two teams within the Trust organization at Airbnb.
• The Foundational Modeling team built machine learning algorithms, signals, and infrastructure to make Airbnb as safe and trustworthy as possible. We shipped step function changes in the integrity modeling capabilities within the organization including wide + deep architectures, entity embeddings, unsupervised and supervised deep feature learning, graph embeddings, and sequence modeling. My team was the co-recipient of the inaugural CFO award for fiscal discipline as a result of the good business enabled and fraud reduced via modeling improvements.
• We also built mission critical ML infrastructure to address the full cycle of machine learning from development, productionization and operation of models and features in production. Reduced MTTD and MTTR for ML failures by ~100x, feature backfilling workflows by 72x, and critical model and data freshness by an order of magnitude.
• I also managed a data engineering team responsible for rearchitecting critical business intelligence using data engineering best practices. Our improvements improved test coverage, correctness, and reliability by significant margins across the organization.

Nov 2017 – Nov 2018 · 1 yr · San Francisco Bay Area

May 2016 – Oct 2017 · 1 yr 5 mos

• Managed a machine learning team, and tech lead for 2 full stack squads of software engineers and data scientists totaling 15 people.
• My ML work on claims risk and behavior detection using on device sensors spawned entirely new lines of business for the company, enabling it to enter the enterprise space. Other key ML improvements increased revenue by X%.

May 2013 – May 2016 · 3 yrs

• Worked on a variety of machine learning, infrastructure, and data products as the 6th technical employee. The majority of my granted patents come from my work during this time. Helped scale the customer base > 1000x, and engineering + data science team 10x.

Oct 2012 – May 2013 · 7 mos

• Bing Whole Page Relevance Team
• Wrote C++ online query time infrastructure for ranking and featurizing entities from Bing's knowledge graph (Satori).
• Created offline pipeline to train and evaluate stochastic gradient boosted tree models for ranking entities from Bing's knowledge graph. Pipeline was used to train some of the first entity ranking models dependent on a wider, richer set of query time, and whole page signals.
• Feature engineering, ranking investigation and debugging for whole page entity relevance.

Mar 2011 – Oct 2012 · 1 yr 7 mos

• PM for the Bing Domains team, specifically Movies, TV, and Shopping.

Jan 2010 – Dec 2010 · 11 mos

• Parallel programming languages and AI Gameplaying Research

Sep 2008 – Jun 2010 · 1 yr 9 mos

• CS227B, General Game Playing, creating AI agents that play arbitrarily defined games in Game Description Language.
• CS107. Taught low-level c, scheme, python and concurrent programming to undergraduate Computer Science students.

Jul 2009 – Sep 2009 · 2 mos

• Implemented a system for query reformulation/expansion and measuring query similarity. Researched and implemented an alternative profile/document ranking scheme.

Jun 2008 – Sep 2008 · 3 mos · Redmond, WA

• Visual Studio