Johannes Wehrstein

I am a Doctoral Researcher in the Systems Group at TU Darmstadt, working on learned database components. My research focuses on leveraging AI to improve database performance, covering areas such as cost estimation, query optimization, advisory systems, and foundational AI research on query-plan representation learning. My long-term goal is to develop a zero-shot database management system that can automatically optimize for new hardware and databases right out of the box.

Since March 2022, I have been a PhD student in the Data Management group of Prof. Binnig. I completed both my Bachelor's and Master's degrees in Computer Science at TU Darmstadt, with a subsidiary in Business & Law. During my studies, I worked on various topics in database and performance modeling, including a novel approach for executing OLTP transactions on GPUs, a learned empirical performance modeling system, and intra-transaction parallelism for TPC-H workloads.

During 2023 & 2024 I was part of Systems Research at Google, where I worked on Foundation Database Models—models designed to generalize across tasks, workloads, and database systems (CIDR’25).