Joe Bradley

Helping to bring Market Intelligence and AI together.

IDC has one of the world's great proprietary data sets--essentially a reified, quantitative understanding of the world's technology use and its value to business. It also has the eyes and ears of the world's business leaders. I'm delighted to be leading AI globally for the company, working on using this data to invent new products and enhance the ones IDG already has.

Starting in 2018, we built the AI research practice at LivePerson, and invented and launched a conversational AI product suite that powered hundreds of millions of conversations every year. In 2022 that led to generative AI. We created products and builder tools that helped >60 brands use generative AI in many millions of conversations, for both employee (assistance) and consumer-facing (chatbot) use cases.

I managed several functions for Nike's Consumer Knowledge organization, handled by the following teams: Data Science, Personalization Science, Testing & Experience Optimization, Audience Selection & Targeting. My group consisted of an array of scientific professionals, science managers, operations & optimization experts, and engineers.

Started recently at Nike building a team charged with personalization and causal inference. It's a terrific combination of deep science questions and challenging automation/platforming problems. Looking for people who like to build things and are on the spectrum between science and engineering. Hiring now, so come join the team!

I am the data and analytics team leader for the Amazon search experience, and we're hiring. We're a team of engineers, scientists and analysts, and we use ML to learn which aspects of the search experience at Amazon are working, which are not, and find the best way to build a platform that responds in real time to customer needs. To do this, we're turning petabytes of data into discriminative models aimed at predicting slow connections and likely pages of interest, so we can adapt our service patterns and make the site instantaneously responsive. We're also doing unsupervised learning on billions of query events to develop the right way to categorize our customer's interests, and we're trying to learn the signals our customers give us when they're frustrated with their search process. If you're interested in becoming part of a team using machine learning to solve real pain points for customers and impact millions of people every day, and if you've got expertise along the spectrum of analytics, data science and large scale data development, we'd love to hear from you.

I worked as the lead scientist for the Amazon ad platform's targeting effort. I developed models and modeling infrastructure relevant for audience segmentation. This includes discriminative, predictive modeling, but it also includes generative modeling, aimed at discovering latent behavior patterns and reducing feature dimensionality. I also worked as the technical/development lead for a prototype BI analytics suite combining data from a petabyte scale data warehouse with similarly sized advertising data. This tool is now being further productionalized and is credited with >$10MM in incremental revenue. In these roles I used Python extensively, both locally and in the Hadoop context (streaming map-reduce). I worked regularly on both static Hadoop clusters and Hadoop in the cloud (Amazon Elastic Map-Reduce).

I was the founder and principal investigator of a research program involving Livermore, Stanford, the Carnegie Institute, the Argonne National Lab, and the University of Washington. I authored the funding proposal, designed the experimental program, carried out many experiments and wrote and published our results in the highest quality peer-reviewed physics journals (Physical Review, Physical Review Letters). I also gave several invited talks on the work. The work was novel experimental testing of advanced theories of condensed matter physics. It focused on heavy element and rare earth systems, which range in application from nuclear power to next generation super-magnets, and are still surprisingly poorly understood.

At LANL, I worked closely with chemists, performing x-ray experiments and data analysis that tried to understand f-electron bonding. I designed experiments and analyzed hundreds of data sets, fitting models and performing regressions. I also built a data analysis framework that improved throughput for analysis by ~10x and is still in use today.

I performed finite element analysis for thermal modeling of microstructured devices