On the morning of March 31,2025, the School of Computer Science and Technology of HUST held an academic report in conference room 433 in Nanyi Building, with the theme of  "AI-Native Data Analytics Systems: Past, Present and Future". The lecture, focusing on the current hot areas, was delivered by Professor Lu from the School of National University of Singapore, attracting many teachers and students to listen and learn. Professor Zhang Rui, from the School of Computer Science and Technology, gave an opening speech and introduced Mr.Lu's academic background and research field. In the report, Professor Lu explores how large, heterogeneous data sets can be processed using large language models (LLMs), and the huge computational challenges posed by this process.

Professor Lu pointed out that the task of analyzing 100TB of financial news for sentiment analysis takes more than a decade to run on a single NVIDIA L4 GPU. This phenomenon demonstrates the huge pressure of current data analysis in computational costs, especially in the context of multimodal data analysis, when the computational requirements become more severe. Traditional online analysis processing (OLAP) databases and data warehouses, such as Oracle and Snow, while outstanding in bulk query processing in the field of business intelligence, face the challenges of limited resources, larger data sets and more complex data sources in AI-driven scientific data analysis.

Professor Lu further introduced the next generation of AI native data analysis system, and proposed that the efficiency of data analysis can be significantly improved by integrating the new data lake house architecture and decentralized computing and storage. Through AI-driven query processing and optimization, these systems can effectively reduce costs and improve the accuracy and speed of analysis, especially in the face of increasingly complex data in scientific research.

At the end of the lecture, Professor Lu had an interaction with the teachers and students on the scene. During the question-asking session, students enthusiastically asked questions and had in-depth discussions on the AI-driven data analysis system, privacy and security issues, and future technology development.

This report not only enables teachers and students to have a deeper understanding of the AI native data analysis system, but also provides a valuable academic vision for the future development direction of data analysis technology. The report was successfully concluded in the heated discussion, and everyone said that they had gained a lot.

Introduction of the speaker:

Yao Lu is an Assistant Professor of Computer Science at the School of Computing, National University of Singapore. His research spans AI and data analytics systems, with extensive experience in both academia and industry. Before joining NUS, he spent over a decade as a researcher at Microsoft Research Redmond, collaborating with teams from Bing and Azure Data Platforms. He earned his PhD from the university of Washington.