Trends and Challenges in Industrial Data Systems Research (IDSR) Workshop

in Conjunction with 51st VLDB

Wesley Room, 4th Floor,Queen Elizabeth II Centre (QEII Centre),London, UK

Sep 1st, 2025

Overview

In the past few years, we have witnessed a number of important advances in the industrial data systems landscape, centered around major industry trends such as the emergence of "all-in-one" Lakehouse architectures, the support for new hardware platforms, the predominance of cloud-based systems, and the continued adoption of ML/AI techniques for a wide range of purposes, from advanced analytics and user interfaces to query optimization, resource management and system autonomics. These advances pose a number of important research challenges and opportunities for innovation in the industrial data systems space; these include, for instance, managing the enormous system complexity of this new generation of data engines, providing seamless integration and support for new heterogeneous hardware and accelerator architectures, and effectively exploiting the state-of-the-art advances in ML/AI (such as novel foundation models and agentic AI).

The IDSR workshop aims to bring together leading industry researchers and practitioners to discuss the state of the industrial data systems field, including recent trends, developments and accomplishments, as well as ongoing challenges and future directions for industrial research and systems development. The program will be structured to promote interactive discussion and exchange of ideas across major industry players, through insightful presentations and open discussion sessions/panels. The end goal of the workshop will be to produce an Asilomar-style report on the state of industrial data systems research comprising contributions from all workshop participants.

Agenda

Date: September 1st, Monday,2025

Wesley Room, 4th Floor,Queen Elizabeth II Centre (QEII Centre),London, UK

Please note that while main IDSR participants will be invited by the workshop organizers, the discussion will also be open to anyone who would like to attend (limited, of course, by the room capacity). No advance registration is required.

Time

Session

Topics

8:30-9:00

30mins

Opening Remarks

Welcome & Intro: Organizers

Self-introductions around the table :All

Brainstorming: Workshop structure and agenda : All

9:00-10:00

60mins

Data Systems in the AI Era-I

LLMs and data management, Fatma Ozcan (Google)

LLM-Native data processing system, Bolin Ding (Alibaba)

The AI-first database, Yannis Papakonstantinou (Google)

Discussion & Brainstorming (All)

10:00-10:30

30mins

Coffee Break

-

10:30-12:00

90mins

Data Systems Architecture & Implementation

The New Memory Wall and how it changes database system design, Anastasia Ailamaki (EPFL)

Aurendil: Challenges in graph databases, James Clarkson (Neo4j)

Challenges in Serverless Multi-Modal Data Processing, Kai Zeng (Huawei)

• Challenges in developing RecSys and agentic interactions, Manos Karpathiotakis (Meta)

Discussion & Brainstorming (All)

12:00-13:30

90mins

Lunch Break

-

13:30-15:00

90mins

Data Systems in the AI Era-II

ML and GenAI for systems, Tim Kraska (AWS/MIT)

Metadata is Everything or Everything is Metadata, Michael J. Franklin (U Chicago)

Productizing NL2SQL and data agents in Microsoft, Fotis Psallidas (Microsoft)

Autonomous agents: beyond Q&A, Yiwen Zhu (Microsoft)

Discussion & Brainstorming (All)

15:00-15:30

30mins

Coffee Break

-

15:30-17:00

90mins

Other Topics & Challenges and IDSR Report Sync-up

Discussion & Brainstorming (All)

Closing Remarks

-

-