Paul Martin McNeill
Digital solutions developer and AI consultant · MSc Artificial Intelligence, Queen's University Belfast
Profile
Developer, AI consultant and MSc Artificial Intelligence graduate applying machine learning to real-world water infrastructure data. On a consultancy contract with a UK water utility, I build and maintain models that forecast leakage across more than 1,250 district metered areas, using flow, weather and operational data that are noisy, irregular and incomplete, and producing results that must be explainable to a regulator. I hold a First-Class BSc (Hons) in Computing and am seeking a PhD in machine learning for imperfect, multisource real-world data.
Research interests
Machine learning with incomplete, heterogeneous and multisource data; environmental and infrastructure time series; forecasting under seasonal and regional distribution shift; uncertainty estimation; explainable models that support real decisions.
Education
MSc Artificial Intelligence Queen's University Belfast Completed
- Modules: Foundations of AI; Machine Learning; Knowledge Engineering; Computer Vision; Natural Language Processing; Artificial Intelligence for Health.
- Themed Research Project.
- Studied remotely alongside full-time employment.
BSc (Hons) Computing, First Class The Open University 2025
- Final-year project: an augmentative and alternative communication (AAC) application.
- Focus on artificial intelligence, data analytics and software engineering.
M816 Data Management (postgraduate module) The Open University 2026
- Data governance, data modelling and database optimisation.
A Levels St MacNissi's College, Garron Tower 2008
- Applied ICT (A), Biology (B), Geography (B), Chemistry (C).
Research and professional experience
Digital Solutions Developer & AI Consultant Consultancy contract with a UK water utility Nov 2025 – Present
- Built and maintain a weekly machine learning pipeline forecasting leakage for 1,253 district metered areas (DMAs), achieving R² = 0.965 and MAE = 0.111 l/s.
- Monitor model reliability in operation: diagnosed feature drift, a systematic regional over-prediction bias and changes in the number of assessable DMAs, and added physical constraints to forecasts.
- Built cleaning and harmonisation pipelines for large operational datasets recorded at different resolutions and in different units, and integrated Met Office and Open-Meteo weather data as model inputs.
- Developed a DMA priority score analysis to target leak detection resources across the network.
- Support regulatory reporting using water balance, minimum night flow (MNF) and IWA leakage methods.
Creative Director / Technical Lead Paramount Perceptions (self-employed) Mar 2019 – Present
- Design and build AI-assisted web applications and analytics for clients in healthcare, education and retail, including LLM-based tools and predictive models.
- Manage projects end to end, including GDPR-compliant data handling and communication with stakeholders.
CRM Systems Coordinator Industrial Temps, Belfast Nov 2018 – Mar 2019
- Consolidated CRM data and reporting from multiple sources; trained staff, improving data accuracy by over 25%.
IT & Database Systems Manager Platinum Support & Care Services, Ballycastle May 2016 – Nov 2018
- Designed a central database and Power BI KPI dashboards, reducing administrative workload by 40%.
Earlier roles (2013 – 2014): e-commerce, multimedia and IT design assistant positions.
Independent research
River early warning across 671 British catchments 2026
Reproducible study on CAMELS-GB: forecasting river flow and flood and low-flow warnings 1 to 7 days ahead without future weather; transfer to unseen regions and a drier climate; a regional LSTM against gradient boosting; calibrated forecast ranges. Code.
Drug response beyond a single marker 2026
Reproducible study on GDSC (924 cancer cell lines): how much wider gene data adds to PTEN status, a measured feature-selection leak, and the stability of clusters and explanations. Code.
Protocol fixed before any model was fitted, baselines first, fingerprinted splits, a public log of every decision and correction. Built making use of AI tools under research rules I set; re-run on my own hardware. Not yet peer reviewed.
Technical reports
- Interim freeze-thaw preparation report, winter 2026/27. Author. Technical report for a water utility.
- Long-term leakage strategy to 2050. In preparation. Contributing author; produced figures through a reproducible document build pipeline.
Selected projects
AI-Powered AAC App assistive communication for non-verbal users 2024
- Built end to end a mobile and web app for non-verbal users, with on-device next-word and phrase prediction (an offline n-gram model with an opt-in personal layer and a small reranker; each suggestion shows why it was offered) that personalises with use, pictogram vocabularies and camera-based scene captioning.
- Evaluated through weekly pilots with users and caregivers: 35–50% faster time-to-utter for common phrases (pilot, n = 7) and +40% caregiver satisfaction with session clarity.
- Designed for WCAG 2.1 AA accessibility and GDPR-first privacy (anonymised telemetry, consent flows, data export and deletion), with a caregiver analytics dashboard; React Native, React, Firebase.
Aurora, connected-care reporting
- End-to-end ETL and Power BI reporting on visits, workforce and finance, built on synthetic NHS data.
AI-enhanced social care management
- Client management system with an NLP assistant, predictive analytics and safety monitoring.
Technical skills
Machine learning: Supervised and unsupervised learning, time series forecasting, model evaluation and monitoring (drift, bias), interpretability; scikit-learn, TensorFlow, PyTorch, Hugging Face, MLflow.
Data: Python (pandas, NumPy, matplotlib, Jupyter), SQL (PostgreSQL, MySQL), MongoDB, ETL pipelines, Power BI, weather APIs.
Software and cloud: Git/GitHub, Docker, Kubernetes, FastAPI, Flask, TypeScript, React, Java; AWS, Azure, Google Cloud (Vertex AI); LLM APIs.
Certifications and memberships
- Google Cloud Professional Machine Learning Engineer (2024; recertifying December 2026)
- AWS Certified Machine Learning, Specialty (2025)
- DeepLearning.AI TensorFlow Developer (2024)
- PCPP1 and PCPP2 Certified Professional in Python Programming (2024)
- MBCS (British Computer Society)
- IET
Volunteering
Digital Volunteer RSPB Northern Ireland, Rathlin Island Project 2024 – Present
- Help develop Rathlin360.com, a digital platform on the island's biodiversity, conservation and heritage, working with conservation and education teams.
References available on request.