Ciara Donnelly
Founder and lead ML engineer
Former data science lead at a Belfast logistics firm. Ciara designs model architectures and owns technical quality across all projects. She holds an MSc in Machine Learning from Queen's University Belfast.
How we got here
In 2019, our founder Ciara Donnelly was leading the data science team at a Belfast logistics company. She kept seeing the same pattern: vendors would sell impressive demos, then deliver models that broke the moment they touched real data. The gap between AI research and production deployment was enormous, and nobody seemed interested in closing it.
She left to start Ai Optimal Solutions with one goal: build AI that survives contact with messy, incomplete, real-world data. The first client was her former employer. The second was a referral from that client. Word spread.
By 2021 the team had grown to five. Today we are nine, and we have shipped 47 production systems across logistics, insurance, retail and manufacturing. We still work from our office on Torp Fields in Smithamford, though most client meetings happen on-site at their facilities or over video.
The team
Small enough that every person knows every project. Large enough that nobody is a single point of failure.
Rajan Patel
Senior data engineer
Rajan builds the pipelines that move data from client systems into our training and serving infrastructure. Before joining in 2020, he spent six years at a London fintech company working on real-time transaction processing.
Fiona McBride
ML engineer
Fiona specialises in computer vision. She trained the shelf-stock detection model we deployed for a UK grocery chain and the safety-helmet detection system used on three construction sites in Belfast.
Tomás Gallagher
Project manager
Tomás keeps timelines honest and clients informed. He manages four to six concurrent projects and runs the weekly stand-ups with each client team. His background is in software delivery at Kainos.
The rest of the team: two more ML engineers (Aoife and Darren), a second data engineer (Priya), a UX designer (Liam) who builds our monitoring dashboards, and a second project manager (Sinéad) who joined in early 2024.
Key milestones
2019 — Founded
Ciara registers the company and signs the first client within six weeks. The project: a demand forecasting model for a cold-chain logistics provider that reduced over-ordering by 18%.
2020 — Team of five
Rajan and Fiona join. We deliver our first computer vision system: a quality inspection pipeline for a food packaging plant in Craigavon that catches defects human inspectors miss 12% of the time.
2021 — Managed AI launch
We introduce the Managed AI plan after three clients ask for ongoing model maintenance. Monthly recurring revenue lets us invest in better tooling, including our internal retraining orchestrator.
2022 — 25 production systems
Milestone project: a document intelligence pipeline for an insurance broker that extracts claim data from 14 different form layouts. Processing time drops from 22 minutes per claim to under 90 seconds.
2024 — Team of nine, 47 systems shipped
Sinéad joins as second project manager. We begin working with our first clients outside Northern Ireland, including a retail chain headquartered in Manchester and a manufacturing firm in Edinburgh.
What we believe
These are not wall posters. They are the rules we use to make decisions when a project gets difficult.
Ship working software
A model that runs reliably at 85% accuracy is more valuable than a prototype that hits 92% in a notebook and crashes in production. We optimise for deployment stability first, then improve accuracy incrementally through retraining.
Say no early
If your data does not support the use case, we tell you in the Assess phase. Spending £2,400 to learn that is cheaper than spending £15,000 to confirm it.
Own the outcome
We do not hand off a model and disappear. Every Build project includes 90 days of post-launch support. If something breaks, we fix it. If accuracy drifts, we retrain.
Keep it explainable
Your team needs to understand why the model makes the predictions it makes. We document feature importance, provide example-based explanations and build dashboards that surface the reasoning, not just the output.
Get in touch
We are happy to talk through your project before you commit to anything. Call, email or fill in the form on our home page.
32 Torp Fields, Smithamford, Northern Ireland, WP97 1TR, United Kingdom