Building the future of privacy-enabled AI

Partnered with a wide variety of internal and external customers developing critical ML projects. I was generally someone that got tapped early to try to get something to work, and if I was able to show it was profitable/tractable Salesforce would put a team together for it. This happened multiple times, from data science for search engine optimization, to predictive ML for infra/devops automation, to service automation through large language models. Internal SFDC Projects: Predicted testing failures with novel DNN techniques, identified team assignments for application errors and failures, and identifying graph-oriented structural inefficiencies in module-linking patterns for our monolithic SFDC application. Created high-precision forecasting models for employee workload and case volume. Completely overhauled the search ranking pipeline at SFDC, adapted legacy code, introduced/cleaned loglines, introduced A/B tests for ranking, analyzed and identified optimal coefficients for ranking models that added no load on legacy systems.SFDC Customer Projects: Created a high performance anomaly detection system at Hulu, candidate matching systems at Allegis/CPL, and a proactive customer intelligence/case deflection system at Citibank.

I am a co-instructor for ML 410 - "Applied Machine Learning"​ in the spring of 2017. I've made slides from the course available here: https://github.com/jdonaldson/uw-mlearn410I also served on an advisory board for course/curriculum design.

I co-founded BigML with the desire to make Machine Learning on Big Data a practical option for more companies. At BigML we believe we are the next step for business analytics. Our systems do not just provide summary statistical reports of past data, but provide predictions on likely outcomes of events in the future.

Thesis: Analysis of music relational information sources for visualization, navigation, and discovery Also: Research and studies in interactive visualization, multi-dimensional scaling, user experience, human-computer interaction, interaction design, social network analysis, and music signal processing.

I worked as an intern, researcher, and developer at MusicStrands/MyStrands/Strands.com for several years. I created several interactive applications and visualizations focused on multidimensional scaling, network layout, using web based interaction design. Developed working prototypes and products in a range of web oriented languages and technologies.Authored 2 patents on related visualization techniques.Hosted a summer of code event for the Haxe programming language.Example Project Strands "tour 3d" : Enabled Strands.com's fitness users to view a 3d replay of their workout route. Created a framework for custom camera tracking and control code in javascript. Created 3d models compatible with Google Earth, and all other media assets for the project itself.