Graph databases have revolutionized the way we model and query highly connected data, but scaling them for read-intensive analytical workloads poses unique architectural challenges. This talk will introduce Neo4j and the Labeled Property Graph model, exploring how query languages like Cypher perform complex pattern matching while addressing the inherent difficulties of working with dynamic, real-world graphs.

Building on this foundation, we will introduce Project Aurendil, a next-generation graph query engine, framing its architecture through the lens of graph transformation. In Aurendil, query optimization is not treated as a static pipeline of relational operators, but as a dynamic graph rewriting problem. We will explore how Aurendil lowers Cypher queries into a “Sea-of-Nodes” graph—a highly specialized Intermediate Representation (IR). By applying continuous graph transformations to this IR—guided by dynamically changing metadata about the graph schema—the engine progressively translates queries from abstract pattern-matching rules directly into highly optimized, native execution binaries.

Crucially, we will highlight how this leap in performance directly empowers the modern AI ecosystem. As Graph Retrieval-Augmented Generation (GraphRAG) and autonomous AI agents become the standard for enterprise knowledge systems in 2026, they require near-instantaneous, multi-hop reasoning over massive datasets. We will discuss how this intersection of database systems engineering and graph transformation principles allows Aurendil to deliver the massive speedups necessary for complex AI and analytical workloads, establishing the next-generation graph runtime as the structural backbone for AI reasoning.

Short CV and Bio

Dr. James Clarkson is a UK-based industrial researcher working to inspire the next generation of cloud-based graph databases. Supported by a UKRI Future Leaders Fellowship, he oversees the Aurendil project at Neo4j Research, where he and his team are developing forward-thinking technologies to underpin the emerging graph-based AI economy. Before joining Neo4j, James received his PhD from the University of Manchester in 2019 for the creation of TornadoVM. He brings a wealth of cross-disciplinary expertise to his current database research, having previously worked in cybersecurity, High-Performance Computing (HPC), and robotics for industry leaders including ARM and Dyson.

Dr. James Clarkson is a UK-based industrial researcher working to inspire the next generation of cloud-based graph databases. Supported by a UKRI Future Leaders Fellowship, he oversees the Aurendil project at Neo4j Research, where he and his team are developing forward-thinking technologies to underpin the emerging graph-based AI economy. Before joining Neo4j, James received his PhD from the University of Manchester in 2019 for the creation of TornadoVM. He brings a wealth of cross-disciplinary expertise to his current database research, having previously worked in cybersecurity, High-Performance Computing (HPC), and robotics for industry leaders including ARM and Dyson.