Bryan Tower
Principal Software Engineer
About
Bryan Tower is a Principal Software Engineering Manager at Microsoft, on the Agentic Data Memory team and driving innovation at the intersection of retrieval systems and applied machine learning. His recent work includes producing a source‑code–optimized version of GraphRAG (opens in new tab) that reduced memory usage by more than 80%, improved performance by 70%, and is now deeply integrated into internal engineering workflows.
Bryan’s applied research spans petabyte‑scale data infrastructure, graph analytics, large‑scale machine learning, and information visualization. He develops systems that uncover patterns in massive datasets and has contributed to projects in neuroscience, cybersecurity, counter‑human‑trafficking, fraud analytics, information retrieval, retail analytics, and real estate.
Previously, Bryan’s team improved the scalable vector indexing methods that enable large language models to perform high‑quality semantic search. They help power DiskANN (opens in new tab), Microsoft’s high‑performance approximate nearest‑neighbor system for large‑scale vector data, supporting real‑time updates and filters while maintaining strong accuracy and cost efficiency.
His past work in network machine learning includes analyzing graphs with billions of edges and contributing to Microsoft Research’s Studies in Pandemic Preparedness initiative. This research influenced organizational science and was featured in the Harvard Business Review (opens in new tab). The team’s network analytics also supported Bing recommendation engines, anti‑fraud systems, and news provenance analysis. Many supporting libraries were open‑sourced with Johns Hopkins University as the graspologic (opens in new tab) project.
Bryan has developed data‑science tooling at the intersection of AI and business intelligence, contributing text‑analytics capabilities to Power BI and shipping multiple open-source Power BI visuals, including Network Navigator, Time Brush, Table Sorter, Attribute Slicer, Cluster Map, Facet Key, and the Strippets Browser. He also delivered solution templates for social brand management and news analytics.
He has applied machine learning to protect users from phishing fraud at global scale and has helped multiple teams detect and stop fraudulent behavior.
Before joining Microsoft, Bryan led data‑science and big‑data efforts across commercial and government programs. His work included revenue prediction and optimization for major retailers, workflow and search‑experience improvements for the U.S. Army, and large-scale graph‑analysis initiatives. He also led a team that won a USDA innovation challenge competition.
Bryan holds a graduate degree specializing in Artificial Intelligence and Information Retrieval.