Jonathan Larson
As of October 2026, Jonathan Larson is a Partner Data Architect at Microsoft Research and leads a research team on graph machine learning, LLM memory, and orchestration. He studied computer engineering at Washington State University, led research and development at Applied Technical Systems, served as Chief Scientist and Technical Fellow at Sotera Defense Solutions, and led data science efforts at Google, Zillow, and the US Army. In 2020, he coauthored graph embedding work on query understanding and its transfer to Bing. His 2024 GraphRAG and LazyGraphRAG work developed graph-based answers to questions over private datasets, and Microsoft added LazyGraphRAG to Discovery and Azure Local services in June 2025.
- Now at
- Microsoft Research
- Works
- 41
Publication numbers from OpenAlex, as of 15 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.
Timeline
-
2000–2015
Computer engineering and early roles
He studies Computer Engineering at Washington State University from 2000 to 2005 and holds BS and MSc degrees. He directs research and development at Applied Technical Systems from 2006 to 2010, then serves as Chief Scientist and Technical Fellow at Sotera Defense Solutions from 2011 to 2015 on assignment to DARPA. His data science work also includes Google, Zillow, and the US Army.
-
Since 2015
Leading a graph research team
His Microsoft Research role begins in 2015. He leads a team focused on graph machine learning, LLM memory representations, and LLM orchestration.
-
2020
Graph embeddings reach Bing
In 2020, he and coauthors present interfaces for exploring graph embeddings built from search-engine query logs. Their paper describes how visualization supported transfer to Microsoft's Bing team and its decision to ship the technology on bing.com.
-
February 2024
GraphRAG for private data
In February 2024, he and Steven Truitt introduce GraphRAG, which uses LLM-generated knowledge graphs to improve question answering on complex private documents.
-
April 2024
A method for global questions
In April 2024, he coauthors a GraphRAG paper that builds an entity knowledge graph and community summaries for answering global questions over private text collections.
-
November 2024
Introducing LazyGraphRAG
In November 2024, he introduces LazyGraphRAG with Darren Edge and Ha Trinh, a graph-enabled RAG method that defers LLM use until query time.
-
June 2025
LazyGraphRAG enters products
In June 2025, Microsoft integrates LazyGraphRAG into Microsoft Discovery and makes it available through Azure Local services in public preview.
-
As of October 2026
Partner Data Architect at Microsoft
As of October 2026, he is a Partner Data Architect at Microsoft Research and leads a team focused on graph machine learning, LLM memory representations, and LLM orchestration.
Sources and credits
Photo: We have no published photo of Jonathan Larson yet. Send us a free-licensed photo.
- Jonathan Larson | alphaXiv
- Jonathan Larson at Microsoft Research
- Making Sense of Search: Using Graph Embedding and Visualization to Transform Query Understanding
- GraphRAG: Unlocking LLM discovery on narrative private data
- From Local to Global: A Graph RAG Approach to Query-Focused Summarization
- LazyGraphRAG: Setting a new standard for quality and cost