Data systems · AI4DB · DB4AI

Songsong Mo

Research Fellow, College of Computing and Data Science
Nanyang Technological University, Singapore

My research focuses on AI for database systems and database systems for AI, especially query optimisation, semantic query processing, and heterogeneous and multimodal data systems.

Email Google Scholar DBLP

About

I am a Research Fellow at NTU's College of Computing and Data Science. I design algorithms, build database prototypes, and evaluate them under realistic workloads. My work has appeared in ACM SIGMOD, PVLDB, IEEE ICDE, IEEE TKDE, and KDD.

I received my PhD in Computer Science from NTU in March 2026. My doctoral research examined how AI can improve query optimisers while retaining the guarantees and engineering discipline expected from a database system.

Research focus

Semantic Query Processing

Execution and optimisation for queries that combine relational operators with semantic operators backed by learned models.

Heterogeneous and Multimodal Data Systems

Storage, indexing, and query processing across relational, vector, spatial, text, and image data within the same workload.

Query Optimisation and Database Internals

Learning-enhanced and parametric optimisation for concurrent workloads, implemented and evaluated in database systems.

Read about my research programme

Selected systems and projects

Current research

Semantic Query Processing

I am developing execution and optimisation methods for workloads that place model-backed semantic operators in the same plans as conventional relational operators. The work studies database-kernel execution and plan selection when model evaluation is expensive and latency, resource use, and result quality all matter.

Ongoing research; further details will be added as results become public.

Open-source system · ICDE 2026

ARCADE

ARCADE is a real-time data system for hybrid and continuous queries over vector, spatial, text, image, and relational data, built on RocksDB and MySQL. I led the design and implementation of its query-optimisation component and contributed to end-to-end system integration.

Paper Code

SIGMOD · PVLDB · Industry collaboration

AI for Database Optimisation

This programme examines how learning can replace selected optimisation logic, enhance an existing optimiser, and transfer optimisation knowledge across tasks. I implemented PostgreSQL prototypes that resulted in lemo, RankPQO, and TATA.

In an NTU collaboration with OceanBase, an enterprise-grade distributed database originating from Ant Group, I led the day-to-day technical work on parametric query optimisation. The resulting ScalePQO work was integrated into OceanBase and reported in a PVLDB industry paper.

lemo RankPQO TATA