Duncan Bell

Duncan Bell
QinetiQ US

Duncan Bell is Principal AI Engineer at QinetiQ US, where he leads the AI & Advanced Analytics capability domain within the Data and Technology Group, QinetiQ’s internal tech incubator. Duncan has operated in the data space for more than a decade and is passionate about solving complex problems through automation and intelligent application of data. As a data scientist, Duncan’s area of focus has primarily surrounded the field of Natural Language Processing (NLP).
Duncan’s largest body of work occurred at the US Marshals Service within its Judicial Security Division; Open Source Intelligence (OSINT) unit, which is charged with the protection of the Federal Judiciary. There, Duncan revolutionized the unit’s existing intelligence gathering & analysis workflow by developing an automated threat screening application called ThreatCanaryAI. ThreatCanaryAI employs a fleet of webscrapers and a trained language model to dramatically expand an intelligence analyst’s capabilities, both in terms of coverage and throughput. ThreatCanaryAI improved unit referral productivity by over 200%, and its predictive model was 42% more accurate than competitive third parties like Babel Street. For this work, Duncan received the Innovation Award at QinetiQ’s global recognition gala in 2023.
More recently, Duncan has developed a Large Language Model (LLM) powered app called PINKY (Proposal Intelligence Neural Knowledge Yielder), which is designed to accelerate the path to a rough draft of a proposal in response to an RFP. PINKY leverages a powerful framework called Retrieval Augmented Generation (RAG), which enables an untrained open-source language model to produce high quality responses.
Prior to QinetiQ, Duncan was Vice President of Data Services at Columbia Books & Information Services, where he led company data collection, governance, and analysis efforts, building and maintaining data pipelines from various public sources. In this role, Duncan designed a novel data matching algorithm which became the basis for a CRM data hygiene application called Bumblebee.
Duncan received his bachelors of political science from James Madison University.