About our Artificial Intelligence practice
We are a small, focused consultancy that builds machine-learning systems for companies that need results, not slide decks.
How we started
Dynamic Ai Frame began in 2019 when two machine-learning engineers at Queen's University Belfast kept getting the same request from local businesses: "We have data, but we don't know what to do with it." The gap between academic research and practical deployment was wide, and most consultancies quoted six-figure budgets before anyone had even looked at the data.
So we started taking on small projects: a churn model for a subscription-box company, a document classifier for a solicitor's office, a yield predictor for a mushroom farm in County Armagh. Each project taught us something about what works outside a lab. By 2021 we had a full-time team and a proper office on Old Military Road.
Today we work with clients across the UK and Ireland, but our headquarters remain in Northern Ireland. We like being close to the universities that produce strong graduates in data science, and we like the pace of life here. It helps us think clearly about problems that deserve careful thought.
What guides us
Our mission is to make machine learning accessible to mid-sized businesses that can benefit from it but lack the in-house expertise to build and maintain models themselves.
Honesty about feasibility
If your problem is better solved with a spreadsheet formula, we will tell you. We have turned down projects where the data was too thin or the expected ROI did not justify the engineering cost. That honesty earns trust, and trust leads to longer relationships.
Reproducibility
Every experiment we run is logged with its parameters, data snapshot and random seed. When a client asks "why did the model change?", we can trace the answer back to a specific commit and dataset version.
Simplicity first
A gradient-boosted tree that your team can inspect and debug is often more valuable than a deep neural network that nobody understands. We reach for complexity only when simpler methods have been tested and found wanting.
Knowledge transfer
We document everything: model cards, API specs, runbooks. At the end of an engagement your engineers should be able to retrain, redeploy and troubleshoot without calling us. We want to be invited back because we did good work, not because you are locked in.
The team
Ciara Donnelly
Co-founder and lead engineer
PhD in statistical learning from Queen's University Belfast. Ciara designs model architectures and oversees production deployments. She has a particular interest in time-series problems and spent three years at a London fintech before co-founding the company.
Declan Murray
Co-founder and data strategist
MSc in computer science, previously a senior data engineer at a Belfast health-tech start-up. Declan handles data pipelines, infrastructure and client discovery workshops. He is the person who figures out where your data lives and how to get it into a usable shape.
Amir Hussain
ML engineer
Amir joined in 2022 after completing his MSc at the University of Edinburgh. He focuses on NLP and has fine-tuned models for legal document review, customer-support routing and multilingual sentiment analysis.
Milestones
2019 — First client project
Built a churn-prediction model for a Belfast subscription retailer. The model identified at-risk customers 14 days before cancellation, giving the retention team time to act.
2020 — Incorporated and hired
Registered as a limited company in Northern Ireland. Brought on our first full-time ML engineer and moved into a shared office space on Old Military Road.
2021 — NHS pilot
Partnered with a local NHS trust to prototype a triage-prioritisation tool for radiology referrals. The pilot reduced average wait time for urgent scans by 18%.
2023 — Cross-border expansion
Signed our first clients in the Republic of Ireland and Wales. Opened a part-time hot-desk in Dublin for client meetings.
2024 — MLOps practice launched
Added a dedicated MLOps service after noticing that many clients struggled to keep models healthy post-deployment. This service now accounts for roughly a third of our revenue.