Speaker Bio: Immanuel Trummer is an associate professor of computer science at Cornell University. His research focuses on making data analysis more efficient and more user-large language models. His papers were selected for “Best of VLDB”, “Best of SIGMOD”, and the CACM Research Highlight Award. He received an NSF CAREER grant for his work on database tuning via LLMs and multiple Google Faculty Research Awards. He is also the author of the book “Data Analysis with LLMs”, now available in five languages.

Abstract: Recently, a new generation of large language models (LLMs), combined with powerful frameworks such as Claude Code, has made it practical to generate large code bassuch as C and C++. In my talk, I will discuss several recent and ongoing projects in the Cornell Database Group, aimed at leveraging LLMs for synthesizing highly efficient code for data processing. In particular, I will focus on GenDB, a multi-agent system that generates code for processing complex SQL queries that is specialized to the particularities of the target database, workload, and hardware. The resulting code often outperforms a variety of classical database systems by significant margins. I will conclude my talk by discussing open problems and avenues for future work.

Speaker Bio: Weston is an open source maintainer and a software engineer at LanceDB. Weston has worked on most aspects of data pipelines, from file formats, filesystems, anddatabases and more. At LanceDB Weston is building a data storage system that optimizes columnar file formats for common ML workloads. He is a PMC in Apache Arrow and Substrait and believes in using open standards and open software to build fast data pipelines. LanceDB has embraced agentic development in the past few years, both within the company, and in the form of agent-driven open source contributions. Weston has been at the intersection of developing software for AI workloads and managing software developed by AI agents. He believes strongly in the composable data systems vision and sees a bright future for these systems in the coming age of AI.

Abstract: The rise of machine learning and agentic development has fundamentally reshaped the challenges facing database developers. These workloads have different query patblending OLAP workloads with search and indexing access patterns more commonly associated with OLTP. The users themselves are often now agents, requiring a new user experience, a heightened focus on API development, and more robust safeguards and boundaries. Finally, the development of these databases is often assisted by, if not entirely created by, coding agents. This is especially true in open source development, where contributions are accelerating across the ecosystem. These forces combine to bring new challenges to database systems, from architecture to API, but also new opportunities, with LLMs potentially accelerating feature development in many ways.

Composable data systems are ideally positioned to take advantage of the new environment. Efficient, narrowly scoped components with clearly defined API boundaries and well-known common interchange structures are exactly the kind of building blocks that most efficiently accelerate agentic development. In this talk we will review the opportunities and challenges that we’ve encountered when integrating AI into an open source database project. This will include architectural challenges such as a well-defined plugin architecture and clearly defined stability boundaries. We will review business challenges, such as balancing enterprise value and open source software in a world where users can rapidly prototype their own systems. Finally, we will discuss the development challenges we face as maintenance needs shift ever further from development to review and design.

Speaker Bio: Benjamin is the CEO at Firebolt. He works with the teams on building the fastest and most cost-effective analytical query engine for engineers and agents. Recently, much of that work has focused on synthesizing full database components with coding agents. Benjamin first fell in love with database systems while studying computer science at the Technical University of Munich.

Abstract: Coding agents are reshaping how databases are built. We already see people starting to synthesize peripheral elements of their data stack: connectors, pipelines, pspecialized solutions. Database kernels are the natural next step. What used to require scarce resources of database internals experts is becoming possible for wide audiences of developers. But synthesizing core database components is scary; there need to be strong guarantees around correctness, transactional consistency, performance, and scalability. This is where harnesses play an outsized role. What are the traits of such harnesses, and how can database systems be composed in this new world?