Garrett Eastham

Garrett Eastham
Artificial Intelligence + Digital Commerce

I am a practicing data scientist and serial entrepreneur working at the intersection of Artificial Intelligence and Digital Commerce. I have been working within enterprise ecommerce for the past 6 years - working exclusively with leading retail technology innovators such as Edgecase (which I founded), Bazaarvoice, and RetailMeNot.



Applied Machine Learning

Familiarity core machine learning principles and applying them with standard frameworks (SparkML, TensorFlow, MXNet) and custom development

Web & Customer Analytics

Developed 6 proprietary web analytics systems over the past 6 years as well as analyst proficiency with modern platforms (Google Analytics, Adobe Omniture, IBM Coremetrics)

Apache spark

Develop primarily in Scala leveraging all parts - SparkML, SparkSQL, GraphX, Streaming

Artificial Intelligence

Working in Clojure / Scala to develop systems that model cognitive processes (Decision Making, Visual Perception)

Modern Web Development

Build tools and prototypes in Javascript / Node (usually MEAN stack) and use both standard (Angular) and emerging (React, ClojureScript) frameworks for front-end development

Product Strategy

Early-stage and platform level strategy definition, planning and execution.

Technical Product Definition

Help bridge product and engineering functions during new product development through technical documentation, project planning, and prototyping.

Amazon Web Services

Modern practicing data scientists toolkit: Hadoop, S3, EMR, Redshift

Work History

January 2016 - Present

Co-Founder & Chief Data Scientist
Edgecase, Inc.

ROLE SUMMARY: After several years working to get the company off the ground, I had the chance to finally step into a pure R&D role as Edgecase’s lead Data Scientist. In this role, I define and envision our organizations research agenda and then work to execute experiments and prototypes against that plan to create valuable intellectual property that can be incorporated into our Product Intelligence Platform.

[-] Machine Curation: Develop technologies that can augment or replace aspects of our human curation workflow through applied machine learning.
[-] Taxonomy Development: Enable the continuous improvement of product taxonomy through automated insights, tasks, and alerts powered by algorithms that monitor changing sources of customer sentiment (i.e. – search strings, reviews, social media).
[-] ROI Measurement: Identify methods / algorithms that can assist in better connecting measured shopper behavior (from search / browse) to specific product data enrichment activities.
[-] Predictive Analytics: Research and develop analytics capabilities that leverage the growing global product taxonomy to understand consumer decision making behavior – and (ultimately) make predictions about future purchasing activities.

[-] Developed and helped prototype the proof-of-concept for the Edgecase Insights offering
[-] Managed the data validation and customer migration process from a legacy web analytics system (Python, MongoDB, Redshift) to a more robust, scalable analytics platform (Clojure, Scala, EMR, Redshift)

[-] Data Processing
— Scala / Spark
— Redshift / Postgres
— MongoDB / DynamoDB / Lambda
[-] Artificial Intelligence / Machine Learning
— SparkML (Word2Vec, Markov Graph Clustering)
— Clojure (core.matrix, duckling)
— TensorFlow / MXNet (CNN, RNN, LSTM)
— FastText
[-] Lightweight Prototyping
— NodeJS / Javascript / Clojurescript
— D3 / AngularJS
— ElasticSearch

April 2015 - January 2016

Co-Founder & Chief Product Officer
Edgecase, Inc.

ROLE SUMMARY: I took over the product officer role as we began to prepare the business for a major shift in product strategy (moving from developing / hosting rich discovery experiences powered by our enriched data to improving the performance of core ecommerce systems by directly integrating our enriched product data). I led the strategy definition and planning that has now become the Product Intelligence Platform – which we officially launched in the spring of 2016.

[-] Led executive team through product strategy “re-definition”
[-] Recruited key members of product staff (including my replacement)
[-] Managed the day-to-day product management activities for core components of the platform re-write

September 2013 - Present

Board of Governors
College of Merchandise - University of North Texas

ROLE SUMMARY: The Merchandising and Digital Retailing program at the University of North Texas is the only college-level academic program focused on preparing the next generation of digital commerce leaders and innovators. In a world where modern retail is transforming at an unparalleled pace, it is critical that we work to ensure our education practices evolve as well. As a member of the board of governors, I work to bring an industry perspective to the program’s academic leaderships.

March 2012 - March 2015

Co-Founder & Chief Executive Officer
Edgecase, Inc.

ROLE SUMMARY: Edgecase (formerly Compare Metrics) began as an extension of my original research on decision interfaces within Stanford University’s Human Computer Interaction program. As I learned more about the enterprise ecommerce market, I saw a huge gap in the tools available for digital merchants to create rich discovery experiences and the product data (and resulting data operations) necessary to power those experiences. As a result, Edgecase was born to bring a scalable SaaS product discovery platform backed by an outsourced BPO model (i.e. – our flexible human curation workforce). I built a large portion of our original MVP, recruited an amazing executive and founding team as well as critical advisors, and raised the initial capital required to get the venture off the ground.

[-] Raised $8 million of venture capital from top-tier VC firms (in Austin and across both coasts)
[-] Recruited key executive leadership positions across sales, clients services, and marketing
[-] Developed initial MVP and worked with our engineering team to get it launched on our initial cohort of customers including Wasserstrom, Crate&Barrel, and Lenovo.

[-] Front-End
— Javascript / Backbone.js /
— Grails / Groovy
[-] Back-End
— MongoDB
— NodeJS / Express
— ElasticSearch

[-] [12/477,831] Method and apparatus for real-time document comparison and feature discovery

December 2010 - March 2012

Technical Product Manager
Bazaarvoice, Inc.

ROLE SUMMARY: My time spent at Bazaarvoice was a transformational moment in my career. I was hired out of Stanford University’s Computer Science program to take over the product management of the company’s existing web analytics integrations (auto-tag functionality that pumped critical user events into pre-integrated systems – Omniture, Coremetrics, and Google Analytics). While there, I developed and managed pivotal research into understanding the impact of user generated content (i.e. – reviews, Q&A, etc.) on shopper purchase behavior through the development of a new analytics measurement technology. This research work not only provided a rich new understanding of the value of social commerce but also allowed me to help jumpstart the company’s first investments into big data with the development of a new internal “web-scale” analytics and data processing platform, codenamed Magpie.

[-] “Magpie” Analytics: Founding product owner for an end-to-end, web-scale analytics system
[-] BV “ViewStream” Analytics: Architected and managed the development & testing of a content impression tracking system
[-] Customer Intelligence: Owned and managed various features of Bazaarvoice’s Customer Intelligence product:
— Text Summarization & Highlights: Managed the development of CI’s initial text analytics capabilities, which focus on automatically identifying and surfacing insights from natural language. Led research initiative in exploring semantic approaches to improving text summarization algorithm.
[-] Web Analytics Integration: Managed the development of new feature improvements to existing web analytics tagging capabilities

[-] Javascript
[-] Couchbase
[-] Infobright
[-] Python / R

[-] [61/585,430] Identifying and Assigning Metrics to Influential User Generated Content