Research

Data systems that connect AI, data and quantum computing.

Build data foundations for quantum computing

How can data systems help quantum computing scale?

My group develops data infrastructure for AI-enhanced quantum error correction and quantum circuit simulation. We connect experimental data, physical context, decoder models and reproducible evaluation.

QEC Model Zoo architecture connecting quantum experiments, data, decoder models and evaluation.
QEC Model Zoo connects heterogeneous experiments, shared metadata, decoder models and evaluation. Figure from QEC Model Zoo · NeurIPS 2026

Discover and generate useful data

Which data should a model learn from, and how can we obtain what is missing?

I develop methods to discover relationships across datasets, identify useful features and generate data under explicit conditions. This connects data lakes with learning tasks, from feature discovery to controllable tabular and time-series generation.

Harpoon diagram showing how condition guidance steers synthetic data generation.
Harpoon guides synthetic data generation using conditions supplied at inference time. View slides

Reuse models and optimise their execution

Which model is suitable for a new dataset, and how can it run efficiently on limited hardware?

I study how dataset–model relationships support model selection and reuse, and how relational databases can execute AI models. My work focuses on reusable abstractions and efficient data movement, memory use and model execution on limited hardware.

TranSQL+ workflow from model computations to SQL and database execution.
TranSQL+ compiles model computations into SQL for execution inside a relational database. View slides

Academic service

Conference reviewing: SIGMOD, VLDB, ICDE, ICML

Journal reviewing: Nature Communications, TKDE, VLDBJ, TODS

Research supported by