Darrell Best joined USC-ISI in the Artificial Intelligence Division in 2019 and has since moved to the Networking and Cybersecurity Division in 2024. As a Senior Research Engineer, he contributes to the development of cutting-edge machine learning and artificial intelligence solutions within interdisciplinary fields, with a focus on applications in cybersecurity, natural language processing, and hardware-aware AI. Mr. Best’s work at ISI is focused on advancing AI-driven methodologies that improve model reliability, interpretability, and effectiveness in high-impact domains.

Over his career, Mr. Best has worked on a variety of AI and ML applications, including natural language processing for psychological and behavioral analysis, hardware-aware AI for FPGAs, and reinforcement learning for strategic decision-making in adversarial environments. He has been actively involved in bridging the gap between research and practical deployment, ensuring AI innovations translate into operational success.

Prior to joining ISI, Mr. Best was a Machine Learning Engineer at QinetiQ US, where he developed automatic target recognition (ATR) systems using deep learning techniques for ground-penetrating radar. He also played a key role in AI-driven object detection and sensor fusion solutions for defense and security applications.

Mr. Best has contributed to multiple research initiatives at ISI, including Sonic Screwdriver, MADEIRA, REDACTED, Danube, Green Sight, and Hawkeye. His work has been instrumental in securing funding and driving forward AI capabilities within ISI.

Mr. Best’s research focuses on artificial intelligence, machine learning, and natural language processing, particularly in the areas of sequence modeling, error correction, classification, generative output, and AI-enhanced decision support. His expertise spans deep learning, transformer-based architectures, and probabilistic modeling.

His current research interests include:

• Natural Language Processing: Developing AI models for text classification, predictive modeling, and psychological profiling to support decision-making.
• AI for Sequence Modeling: Investigating AI-driven techniques for structured prediction, including sequence-to-sequence learning and retrieval-augmented generation.
• Error Correction and Reliability: Enhancing AI methods for identifying and correcting errors in structured data, improving the accuracy of predictive models.
• AI for Decision Support: Applying reinforcement learning and probabilistic modeling to optimize complex decision-making tasks in real-world applications.