About
Rusty Data is a small, senior practice building AI and data models people can actually use — rooted in complexity science, working for people and planet.
Rusty Data is a small, senior AI and data-science practice. We build bespoke models for questions off-the-shelf tools can't answer — agent-based simulation, machine and deep learning, reinforcement learning and complex-systems modelling — applied to data as it actually arrives: irregular, gappy and shaped by human process.
The problems we take on share a shape: a system too tangled to reason about by intuition alone, a decision that matters, and data nobody else wanted to touch. Recent work runs from illegal-fishing enforcement and intensive-care decision support to team allocation and sanitation hardware — chosen, wherever we can choose, for what it does for people and the planet.
Small, senior and hands-on — deliberately so. You work directly with the principal from the first whiteboard sketch to the final handover: no account managers, no juniors learning on your project, no layers between you and the person building the model.
We favour interpretable models, reproducible pipelines and writing things down — and we'd rather tell you a model shouldn't be built than build one that misleads.
Because real data is never shiny. It arrives weathered — oxidised by time, gaps and human process — and the job is not to pretend otherwise but to work with it honestly. The name is a small promise: no lab-bench idealism, no benchmark theatre. Models built for data as it actually is.
Rusty Data is founded and led by Chris McWilliams. Chris holds an MRes and PhD in complexity sciences from the University of Bristol's Bristol Centre for Complexity Sciences, following a first-class degree in theoretical physics at Imperial College London — and, before all that, an art foundation at Camberwell. He still works at Bristol as a Research Fellow in the School of Engineering Mathematics and Technology, developing machine-learning decision-support tools for intensive care.
His work has spanned theoretical ecology — including research on how habitat loss destabilises food webs, published in Nature Communications — and clinical machine learning at the bedside of an intensive care unit. Across all of it the lesson is the same: the hard part of data science is rarely the algorithm; it's understanding the system the data came from.
Mission
Rusty Data exists to point serious technical work at problems that deserve it.
AI and data science concentrate power wherever they are applied. We want them applied to ocean enforcement, ecological recovery, fair institutions and public understanding — not only to ad clicks. Rusty Data is a small, deliberate bet that a technical practice can be both rigorous and principled.
It shapes which projects we take on, and how they're done: interpretable models over opaque ones, communities and domain experts in the loop, honest uncertainty over confident nonsense, and deliverables the client can own and understand long after the engagement ends.
Data science can bring huge benefits to many parts of society, but the cost of hiring a contractor can be prohibitive for smaller and not-for-profit organisations. For the right cause, we're willing to work at rates based on ability to pay. If that's you, please get in touch.
Working principles
The people behind the practice
Chris founded Rusty Data and leads its modelling work — a complexity scientist with 15+ years across interdisciplinary research, AI and data modelling, building bespoke models for real decisions, from intensive-care discharge to ecosystem simulation.
Tamzin brings a writer's and ethnographer's eye to Rusty Data — grounding data work in real human stories and making it legible to the people it affects. She created the Fair Tales project, funded by the Jean Golding Institute.
Research
Rusty Data grew out of an academic life spent studying how interacting parts become behaving wholes.
The practice is anchored in Chris's training at the University of Bristol's Bristol Centre for Complexity Sciences — and he continues to work at the university alongside the consultancy. That dual footing keeps the practice honest: consulting work stays anchored to current research, and research stays anchored to real problems.
Chris's PhD, though badged as complexity science, was really theoretical ecology — how habitat loss and species interactions shape whether an ecological community stays standing. Since then his research has moved between two worlds: ecological modelling, and clinical machine learning for intensive care, where he builds decision-support tools from routinely collected patient data.
Underneath both is one enduring interest: how simulation, networks and machine learning can support real decisions in systems too tangled to reason about by intuition alone.
Selected publications
First-author and collaborative work across ecology, clinical machine learning and systems biology.