Nat joined IBM Research in 1997 after running his own systems management company for 10 years (87-97), and after working with mini-computer software development and sales for the prior 8 years (79-87). He works at home and is associated with the AI Tech Division at the T. J. Watson Research Center in Yorktown Heights, NY. http://www.research.ibm.com/labs/watson/index.shtml
His initial research was in systems management where Nat designed and developed Page Detailer with LeRoy Krueger. Page Detailer is a client side tool to measure and decompose web page performance characteristics. This work received a Research Accomplishment Award and Outstanding Technical Accomplishment Award and generated several patents. The tool is available on alphaworks here: http://www.alphaworks.ibm.com/tech/pagedetailer
or in its community here:
Nat also worked with the Cognitive Toolkit: Agent Building and Learning Environment (ABLE), and Semantic Toolkit: Information Representation Inferencing and Sharing (IRIS) teams for research in empirical, analytic, and semantic reasoning. After contributing to e-Commerce projects using pattern matching agents, he conceived, designed and developed the Parametric Analysis Center (PAC) to collect and analyze high volume time series operational data from complex equipment. This work resulted in a Research asset licensed to the US Army Heavy Brigade Combat Team, and a heavy equipment / mining equipment manufacturer, and was used with a variety of automakers for condition based maintenance research. This work received a Research Accomplishment Award and Outstanding Technical Accomplishment Award. An article on the Parametric Analysis Center is here:
Nat lead the collaborative user experience design team for Investigative Reasoning and Reporting for a government contract. He also worked with the Smarter Energy team, designing and developing a framework to gather and analyze time series operational data from a variety of building management systems.
Nat was the chief architect and technical lead for Collaborative Decision Making (CDM). Nat's work with Steven Ross resulted in a asset named Beyond Discussions allowing teams to deliberate to solve problems and make decisions. This asset is being used by the AMS services organization to conduct customer summits for ideation and strategic planning. This work received a Research Accomplishment Award. An article on Beyond Discussions is here:
Nat was the chief architect and technical lead for Best Fit Expertise helping provide tools and services focusing on "return on expertise" leveraging enterprise expertise to increase revenue through improved product and service offerings, and to grow expertise through talent optimization. He is also helping with visualization of high dimensional data for the Center of Innovation for Visual Analytics (CIVA).
Nat lead a "First of a Kind" (FoaK) engagement with a leading New England financial institution providing Visual Analytics for Governance, Risk and Compliance as part of the Cognitive User Experience research initiative in Cambridge, MA. This work is being moved into production by this company. An example of this asset can be seen here:
Nat lead design and development of the Multi-Channel Visual Analytics Framework culled from the FoaK assets as a tool to assess data from multiple data sources, and provide novel visualizations and insights to assist people with data exploration / understanding.
Nat lead design and development, and was the chief architect and project lead for collaborative decision making leveraging many Watson services to provide cognitive assistance in near realtime. This system provided teams of machine operators with asynchronously derived machine status, and allowed both verbal and typed conversations about procedures to address machine anomalies. The system provided both individual conversation context as well as shared context about the machinery.
Nat is currently leading design and development of an AITech framework to provide access to emerging Research technologies to extend Watson Conversation Services (WCS), leveraging Bluemix services (OpenWhisk, Cloudant, Node-RED). By providing a variety of natural language processing and conversation inference engines, this framework should enable a variety of new types of conversation applications to be developed.