Promoted to role when Sapient acquired AI Labs. Lead N. American data science team; new business development.Led Generative AI engagements for customer service representatives and for a market research firm. Designed forecasting models for a luxury retailer. Advised national radio broadcaster on a data and AI strategy.

A founding employee of the AI Labs N. America practice who built the data science team and acquired cloud/professional services for them to succeed. Deeply involved in new business development. Sample engagements: digital media campaign ROI for a cable TV/mobile provider; forecasts of corporate account balances for a U.S. bank; multiple statistical analysis of surveys by for casualty and health insurers; graph analyses of consumer tastes for a national beverage distributor.Please check PS AI Labs out: http://www.psail.com/

Part of the team of data scientists that engage with clients to understand their IIoT problems, and then design and deploy machine learning algorithms to solve them. Worked with client as diverse as steel milling/processing, transportation/fleet management, and semiconductor manufacture. I engineered deep learning and Bayesian algorithms, and was the team's specialist in visualization. In 2018, the Gartner IoT Magic Quadrant put Hitachi Vantara as one of a very few 'visionary' companies in the IoT space. In 2017, Lumada earned Hitachi Vantara accolades as the #1 IIoT company (by ComputerWorld UK) and the IIoT Enterprise Solution of the Year (by IoT Breakthrough).

Acxiom acquired Allant's TV business (Dec 2015) and my 'Director' title became the Acxiom title 'Principal Data Scientist'. I contributed to the successful launch of Acxiom's anonymous-to-addressable TV service. Predicted impactful national look-alike audience of TV households starting from first-party data seed audiences. Also introduced more modern techniques for look-alike modeling (e.g., k-Nearest Neighbors and Extreme Gradient Boosting) using considerably larger training and test data sets than typical. I also introduced Bayesian inference to TV campaign response rate reporting and analysis.

In July 2015, I was promoted to Director. I assumed responsibilities for defining and progressing towards a single analytics/data science platform within our division of Allant. I was also the manager for our team's other data scientist.I also continued to be the primary individual contributor. In December 2015, I joined Acxiom via its acquisition of Allant's TV business.

I develop classification and predictive models to optimize local and TV national advertising inventory by predicting which households to target and predicting/evaluating pre- and post-campaign effectiveness. The data combines, in a personal privacy-safe fashion, PB-sized databases of: third-party household attributes, set top box return path data, including video ad impressions. I created these working prototypes for Allant on data at TB-scale using HiveQL, Impala and Python machine learning libraries: --- linear programs that assign customer households into test/control experimental groups when proper randomized assignment is critical for actionable campaign measurement --- use statistical extrapolation techniques adapted from social science that start from an a biased/non-representative set of customer households to produce unbiased/representative predictions of national campaign reach. --- train/validate/test different classification methods that predict any pay TV households' affinity for specific movie studio titles. --- quantify the lift TV campaigns have on viewer tune-in to new network programming.

I served as a lead analytics contributor, working with very large data sets and designing tools suitable for web deployment on behalf of FAA policy and budget decision-makers. When prototype tools I helped design were implemented I act as a project manager, accountable for overseeing technical teams of federal staff and contractors.

Implemented algorithms included with software sold by Sabre for trucking, rail and air freight corporations to design and optimize their transportation networks. Methodologies included network optimization, and large-scale linear/integer programming, including various column generation techniques.Served as the lead optimization engine developer for products such as SuperSpin(TM), a flagship network design and optimization software product, and OptiStop(TM), a product designed to be integrated with on-board truck messaging systems that optimized driver fuel stop purchases in near-real time. The original start-up software firm, PTCG, Inc., was purchased by Sabre in 1996.