Overview
The Interfaces research programme extends the AutoAI project at ML@CL. It takes a systems perspective on AI deployment: software is the interface between socio-technical needs and AI model capabilities. The programme targets interpretable, self-sustaining systems where multiple autonomous components cooperate, adapt, and remain accountable in real-world deployments.
Interfaces builds on data-oriented architectures and the Self-Sustaining Software Systems (S4) agenda. Research themes include multi-agent observability, systems engineering for large language models, cooperative intelligence, and methods to approximate understanding of composed system behaviour.
Validation and partnerships
The programme works with the aICU research initiative at Karolinska Institutet and Södersjukhuset, in partnership with ML@CL, to develop and evaluate AI-based decision support for intensive care.
Public open-source artefacts include DOAgent, a library for observable multi-agent systems. Other research prototypes are developed in collaboration with clinical and industry partners.
Programme lead: Christian Cabrera-Jojoa
Related People
Christian Cabrera Jojoa
Assistant Research Professor, Cambridge University
Neil D. Lawrence
The DeepMind Professor of Machine Learning, Cambridge University
Diana Robinson
PhD Student, Cambridge University
Carl Henrik Ek
Senior Lecturer, Cambridge University
Radzim Sendyka
PhD Student, Cambridge University
Andrei Paleyes
PhD Student, Cambridge University
Related Publications
Machine Learning Systems: A Survey from a Data-Oriented Perspective
ACM Computing Surveys, :
Self-sustaining software systems (S4): Towards improved interpretability and adaptation
Proceedings of the 1st International Workshop on New Trends in Software Engineering, :
Requirements are All You Need: The Final Frontier for End-User Software Engineering
ACM Transactions on Software Engineering and Methodology, :
The Systems Engineering Approach in Times of Large Language Models
58th Hawaii International Conference on System Sciences (HICSS-58), :
An Empirical Evaluation of Flow Based Programming in the Machine Learning Deployment Context
1st International Conference on AI Engineering – Software Engineering for AI, :