Projects

Amalur explores the convergence of data integration and machine learning — automating how scattered training data across silos is integrated for downstream models. It is the foundation of the group’s Model Lake vision, where heterogeneous data and rich model zoos meet in one place.

SiloFuse generates cross-silo synthetic tabular data using latent diffusion models, so organisations can share realistic data without ever exposing raw, feature-partitioned records.

WaveStitch performs flexible and fast conditional time-series generation with diffusion models, stitching together realistic signals under user-specified constraints.

TranSQL serves large language models with relational queries — compiling model inference to SQL so that LLMs can run inside a database engine, even on low-resource hardware.

The QEC Model Zoo provides a common benchmark for comparing AI-based quantum error correction decoders with established classical methods. It spans nine real-device datasets and eleven full-decoder methods across multiple code families and hardware platforms.

The associated paper has been accepted to NeurIPS 2026 and is not yet publicly available.

Explore the QEC Model Zoo for an overview of the datasets, decoder families, pre-decoding techniques, and model checkpoints. The codebase will be released when the benchmark manuscript is published.

QEC Lake brings experimental, simulated, and synthetic quantum error-correction data into a shared data lake. Researchers can discover datasets through metadata search and an AI-assisted interface, then export model-ready inputs for decoder training and evaluation.

Its preparation pipeline supports different model families without requiring a separate conversion script for every dataset and model. QEC Lake supplies the data foundation for the QEC Model Zoo and reproducible comparisons across decoders.