Benjamin Squire Email & Phone Number

Las Vegas, Nevada, Us

San Francisco, California, Us

Created pricing strategy for lead generation based on a stochastic multi-armed bandit optimization approach with back end automation and suggestive pricing to enable optimal decisions with respect to budget allocation and ROI. Maintain all BI and Data Reporting across analytics and marketing for the organization, including Front-end Dashboards, client facing access, and all Backend ETL from Data source to Warehouse to Analytics Mart. Python ETL in Apache Airflow with DB backends to Postgres, MySQL, & Snowflake. Migrated entire database warehouse from Periscope/Sisense to Snowflake warehouse and developed all analytic and business reporting queries and dash boards. Reduced operations costs by over 80% through SQL optimization and leaning out analytical needs to critical business functions from legacy systems. Developed scripts in Python which ETL data from LendingTree, analyze optimal bid size and customer filters. Created process to identify optimal pricing strategies and marketing spend in order to maintain competitive pricing and stay profitable. Doubled Customer Transfer rate in first 6 months. Decreased price per lead by optimizing lead filters for customer intake. Documented data intake processes. Developed high visibility dashboards detailing all things lead generation and Marketing. Used Mixed Integer Programming to determine optimal Pricing and Bid amounts for leads.

Des Moines, Ia, Us

Ideated and executed an In-house Enterprise wide Identity Graph of Anonymous and Known Customers across 30+ websites comprising over 18 Billion Cookies and 20 Billion Page views. Provided mechanism to evaluate Data Vendors ability to link anonymous customers using statistics and graph visualizations. ID graph is the main source of customer profiling for 180+ MM profiles across the organization driving retargeting and customer advertising sales.https://go.neo4j.com/rs/710-RRC-335/images/Neo4j-Case-Study-Meredith-EN-US.pdf

Des Moines, Ia, Us

Lead Data Scientist of Identity Platform and Profiling for Graph Analytics. Developed Enterprise wide graph DB in Neo4j capable of syncing ephemeral first party cookies together by means of analyzing how first and third party cookies are related. Developed scalable API for interacting with Identity Graph in python script for use on AWS Lambda. Connecting users across devices and households with first party data for improved advertising targeting and information profiling.Lead Data Scientist of 360 Customer View Initiative, architecting and developing a dashboard platform of cross channel marketing and advertising metrics for Ad Sales and Content Reach performance. First enterprise wide initiative that ties together legacy subscriber and online databases with demographic and NLP enrichment from various sources. Searchable by topic,keyword, demographic, and geographic using Elastic Search. Identifying and designing dashboards around Health-status KPI's of Online properties and verification analysis of multiple engineer's efforts. Providing automated python PSQL scripts which create and load tables for dashboard ingestion on a daily basis. Using Mode Analytics to provide easy access and intuitive wiki based reports that run from AWS EC2 instance.Hackathon Winner GraphConnect2018 - Best Visualizationhttps://neo4j.com/blog/graphconnect-2018-hackathon-stackoverflow/

New York, Ny, Us

Lead Data Scientist of the 360 Customer View Initiative. Helped develop multi-platform, cross-channel dashboard and Machine Learning Capabilities for enterprise wide Ad ops and CM +R. Project was highlighted on Time Inc.'s 2016 10k SEC filing. Senior ML architect who created process design, weekly/ monthly Milestones, implementing AWS ML as elastic automated Look-a-like modeling. Designed evaluation metrics and risk/cost Analysis for Audience Identification. Developing analytics dashboards for business stakeholders across Time Inc. Social and Digital platforms. Provided scalable social listening scripts in Python with the capability to scrap all posts, comments and likes of any public Facebook page and ETL into Neo4j graph database. Developed POC brand and user engagement dashboard with comment word clouds filtered by topics, trending hashtags, and most engaged users analysis. Capable of finding look-a-like users based on user interactions. Lead Data Scientist in evaluation of a major CRM data contributor in comparison with status quo CRM data provider. Developed novel matching algorithm based on unstructured text descriptions to match variables across datasets to enable an apples to apples comparison for number of records delta analysis and model evaluation.

New York, Ny, Us

Creating scalable recommendation algorithms on 20+ mil user datasets for Advertising and Marketing purposes in Python with Web application front end using Flask. Building intuitive web applications using FlexDashboard in R to predict and forecast Ad expenditure across 100's of companies and international brands for effective targeting purposals. Developing graphical roadmaps of Time Inc. data systems and landscape for executive level presentation. Including ER diagrams of SQL systems and join mapping to show how various tables and entities are interconnected.ETL processing social media data such as Facebook into a graph database (Neo4j). Performing POC dashboard analysis across Time Inc. Brands and competitors including topic modeling and user look a like analysis. Developed a database atlas and ETL mapping tool in Python, yEd, and Neo4j showing processes from Oracle datalake to data warehouse to data store.Searchable by transformation, data type, with full explanation of variables that were previously obfuscated in data dictionary of 200+ separate excel spreadsheets.

Vancouver, British Columbia, Ca

On a 4 month project I helped developed and implement an end to end scalable Web application capable of scanning news aggregators and social media for text mining purposes. The title of the project was "Mining Unstructured Text for Risk Assessment". I personally developed multiple Python scripts which allowed structured analysis on unstructured text. In May 2015, I presented some of my summer research in demonstration at Vancouver Analytics meet up Group through meetup.com. I demonstrated scripts I wrote in Python to perform LDA using Gibbs sampling on discourse from the CNN debate show Cross-Fire. The topic was school lunches.In June 2015 I presented at Data driven advancement 2015, my topic was the application and benefits of text analytics in alumni engagement. In October 2015 I presented at the SAS analytics conference as a poster competition runner up. My poster was on "Mining Unstructured text for Risk Assessment".Some of my research touched on topics of machine learning classifiers such as Naive Bayes and SVM as well as LSA and topic Modelling using Latent Dirichlet Allocation.Our work was well received by our client.

Master Of Management, Operations Research

Bachelor Of Science (B.Sc.), Mathematics

Associates In Arts And Sciences, General Studies