AI Powered Tech Scouting & Recommendation Engine

Congress and DoD recognize that small businesses are critical to the US economy and our national defense. Yet DoD spends an astonishing 31.7% of its contracting budget with just 5 large defense contractors and over 63% with the top 100 defense contractors. There are 10M plus small businesses in the US. DoD engages only a tiny percentage of them. Unfortunately, there presently is no effective way for DoD to identify and engage the overwhelming majority of NDCs or to evaluate the value they can provide to DoD or the threat their technology could present if used by a state or stateless enemy. DoD requires a platform to effectively identify, evaluate and engage nontraditional defense contractors that can deliver game changing technology to DoD and the warfighter to ensure capability advantage and mission success.  Third Coast Federal's AI Powered Tech Scouting and Recommendation Engine ("Recommendation Engine") is the solution.   Our existing commercial tech scouting and market research platform, the Federal Navigator, includes a data lake of structured and unstructured data regarding 2M+ businesses, many of which are nontraditional defense contractors (NDCs). Currently, users search for innovators by keywords or Natural Language Processing (NLP) phrases, with the ability to filter by geographic location, NAICS codes and other factors. As good as keyword and NLP search tools are, they are an inefficient and imprecise means for evaluating DoD problem statements and ranking industry capabilities. Searches often return many hundreds, or even thousands, of innovators. As a result, identifying innovators with the precise capabilities required to solve a particular operational or sustainment problem remains a labor intensive process.   For this STTR, Third Coast Federal, Inc. and the University of Notre Dame are enhancing our Federal Navigator tech scouting and market research platform to create an AI Powered Recommendation Engine that leverages knowledge graphs and deep neural nets to instantaneously evaluate DoD needs and simultaneously identify pools of industry and faculty innovators with the technologies and capabilities required to solve identified needs. In Phase I, our team established the technical feasibility of using knowledge graphs and neural networks to produce recommendation algorithms that achieve accurate need comprehension, as well as candidate capability identification, scoring and matching. In Phase II we are developing a prototype Recommendation Engine MVP to help the Joint Artificial Intelligence Center and its constituents engage AI innovators to solve operational and sustainment challenges with AI powered solutions. The MVP solution will employ machine learning and utilize HITL (Human in the Loop) feedback from diverse subject matter experts to provide domain specific refinements to the Recommendation Engine to improve accuracy and performance.