AI & Machine Learning
Production LLM systems — autonomous agents, RAG pipelines, fine-tuned models, and tool-calling workflows wired into real business processes.
AI Engineer · Web Developer · Data Analyst
Building intelligent digital systems
I work at the intersection of AI, web engineering, and data analytics — designing LLM-powered agents, shipping production web apps, and turning messy operational data into decisions that actually move the needle.
Currently
Aurora — NHSCT analytics & ECM transformation
· Power BI suite + Access/SQL data layer · Northern Ireland
15+Years Experience
AI Agents& ML Models
NHSDigital Transformation
Full StackWeb Engineering
What I Do
Three disciplines, one practice
I move fluently across AI engineering, full-stack web, and analytics — so the systems I build don't stop at a prototype. They ship, they scale, and they get used.
Web Engineering
Modern web apps end-to-end — TypeScript front-ends, Node APIs, progressive web apps, and integrations that hold up under real-world load.
Data Analytics
From operational chaos to clean dashboards. ETL, modelling, and reporting that gives leaders something they actually trust to act on.
Selected Work
Projects that shipped
A mix of production systems, client work, and personal builds — each one a decision about what to ship, what to cut, and what to measure.
Reduced reporting cycle from 5 days to under 4 hours
AI + Healthcare
Aurora — NHS Digital Transformation
Northern Ireland-specific, NHSCT-aligned analytics platform — operations, finance, workforce, compliance — built on Access/SQL data layers and Power BI.
Predictive AI assistive communication for non-verbal users
AI Accessibility
AI-Powered AAC App
Mobile + caregiver dashboard that combines on-device prediction, scene understanding, and real-time monitoring for non-verbal users.
Consolidated 6+ disconnected sources into a single live dashboard
Data Analytics
CRM Reporting — Industrial Temps
Data quality framework, ETL via Access + Power Query, and Power BI dashboards for placements, client performance, and consultant scorecards.
Real-time safety monitoring with NLP-driven virtual assistant
AI + Healthcare
AI-Enhanced Social Healthcare Management System
AI-driven platform: client management, NLP virtual assistant, predictive analytics, and real-time safety monitoring for social healthcare delivery.
Natural-language task automation across daily workflows
AI Engineering
AI Assistant
An LLM-powered productivity assistant — fine-tuned models, multiple interface integrations, and real-world automation.
In-development CRM consolidating sales, comms, and analytics
Web Engineering
Custom CRM Web Application
Custom CRM tailored to streamline business operations — sales management, customer communication, and analytics in a single PWA.
Hybrid SQL/NoSQL design with cloud-native storage
Cloud Computing
Cloud-Based Data Management System
Cloud-native database management system using SQL + NoSQL, designed for resilience and elastic scale.
OOP demonstration with full UI, video walkthrough, and source
Software Engineering
Java Ticket Machine
Java OOP showcase — inheritance, polymorphism, encapsulation — with a responsive UI and embedded demo video.
Walkthroughs
See it running
Project demo videos. Click to load — thumbnails first, iframes only on tap to keep things fast.
AI & Machine Learning
The systems I specialise in
Practical AI: agents that take action, retrieval that grounds answers in real data, and predictive models that survive contact with production. Less demo, more deployment.
Autonomous AI Agents
Agents that plan, call tools, and recover from failure — wired into real workflows with guardrails, observability, and human-in-the-loop where it counts.
RAG Pipelines & Knowledge Systems
Retrieval that actually retrieves. Smart chunking, hybrid search, evaluation, and grounding so models answer from your data — not their training distribution.
ML Models & Predictive Analytics
Classical and deep models for forecasting, classification, and anomaly detection — with feature pipelines, experiment tracking, and proper evaluation.
Multi-Agent Orchestration
Specialised agents collaborating across structured workflows — function calling, message passing, and routing logic that keeps complex tasks coherent and debuggable.
In motion
See how an agent thinks
Hand-rolled animation of a research agent — input, planning, tool use, memory, synthesis, output. Auto-plays once when it scrolls into view.
-
User input
"Analyse this quarterly report and summarise the key findings."
-
Planning
Decompose into sub-tasks:
- Extract text
- Identify sections
- Analyse financials
- Detect trends
-
Tool selection
Choose tools for each sub-task:
- document.parse
- nlp.classify
- chart.render
-
Execution loop
Call each tool, capture output:
document.parse(report.pdf)→ 28 pages, 14 tablesnlp.classify(sections)→ 4 categorieschart.render(trends)→ 3 charts
-
Memory
Findings stored in working memory:
Revenue +12% YoY Cost ratio improved EMEA outperformed -
Synthesis
Combine findings into a coherent narrative; check against the original prompt.
-
Output
"Q3 revenue grew 12% YoY driven by EMEA. Margin improved 180bps as cost ratios fell. Forecast a softer Q4 on currency drag."
Live Lab
Try a model in your browser
A small CNN trained on movie reviews, running entirely on your device — no server, no upload. Type a sentence and the model classifies its sentiment in milliseconds. Lazy-loads when this section enters view.
Awaiting input — try a sample or type your own.
Under the hood
- Tokenise — split your sentence on whitespace and look up each word in a 20k-word vocabulary.
- Encode — map words to integer indices, pad or truncate to length 100.
- Inference — feed the sequence through a small CNN (embedding → conv → global max pool → dense).
- Decode — output is a single value 0–1 — closer to 1 is more positive.
Model and vocabulary are fetched from the public TensorFlow.js model host on first use, then cached by your browser. No data leaves your device. The whole forward pass runs on your GPU via WebGL.
Stack
What I'm fluent in
Tiered honestly. No percentages — they don't mean much across different tools and contexts. This is what I reach for, what I ship in, and what I'm familiar with.
Daily driver
Tools I reach for first.
- Python
- TypeScript
- React
- Node.js
- Power BI
- Pandas
- Git
- OpenAI API
- Claude API
Strong working knowledge
Comfortable shipping in production.
- TensorFlow
- Scikit-learn
- LangChain
- Next.js
- PostgreSQL
- Firebase
- Docker
- Azure
- AWS
- DAX / Power Query
Familiar
Used in projects, not my first pick.
- PyTorch
- Hugging Face
- MLflow
- Kubernetes
- GCP / Vertex AI
- Jupyter
- Java
- C++
How I Work
Principles, not posters
How I actually approach work — not slogans on a wall. These are the things I push back on when corners get cut.
-
01
Ship, then iterate
Production-ready first, refine after. Notebooks are for experiments — they aren't a deliverable.
-
02
Measure everything
Monitoring, logging, and drift detection are part of the build. If you can't measure it, you can't improve it.
-
03
Ethics by default
Fairness checks, explainability, and responsible-AI practices are built in from day one — not bolted on under deadline.
-
04
Human-centred
I test with real users, prioritise accessibility, and design for the person at the other end — not the algorithm.
About
Building from Northern Ireland
I'm Paul — an AI engineer, web developer and data analyst based in Northern Ireland. My work sits where engineering meets evidence: shipping LLM-powered tools, modern web platforms, and analytics that hold up to scrutiny.
Recent work includes NHS digital transformation with the Northern Health and Social Care Trust — Power BI suites for operations, finance, workforce and compliance — and AI accessibility tools like a predictive AAC app for non-verbal users with a real-time caregiver dashboard.
On the analytics side, I've rebuilt CRM reporting pipelines, run data-quality frameworks, and turned brittle exports into versioned datasets that survive a Monday morning.
Currently learning
Let's Talk
- M816 Data Management — Open University in progress
- Multi-Agent Systems with CrewAI experimenting
- Kubernetes for ML Workloads learning
- Vector databases & retrieval at scale learning
Building in Public
Recent activity
Live commit graph from github.com/mcneillium. It updates daily — proof I'm still shipping, not just presenting.
Credentials
Certifications & memberships
A record of formal recognition across machine learning, software engineering, and professional bodies — kept current and continuously expanded.
-
Google Cloud Professional ML Engineer
Google Cloud · 2025
-
AWS Certified Machine Learning — Specialty
Amazon Web Services
-
TensorFlow Developer
DeepLearning.AI / Coursera
-
PCPP1 — Certified Professional in Python
Python Institute
-
MBCS — Member of the BCS
British Computer Society
-
IET Member
Institution of Engineering & Technology
Trajectory
A working career
Roles, study, and certifications grouped by era. Compact cards rather than a long stretch.
2024 — Present
Volunteer
Digital Volunteer / Technology Advisor
RSPB Rathlin Island Project
2024 — Present
Role
Creative Director / Technical Lead
Paramount Perceptions (Self-Employed)
2019 — Present
Education
M816 Data Management
The Open University
Expected June 2026
Certification
Google Cloud Professional ML Engineer
Google Cloud
2025
Education
BSc (Hons) Computing — First-Class Honours
The Open University
20252016 — 2018
Role
CRM Systems Coordinator
Industrial Temps
2018 — 2019
Role
IT & Database Systems Manager
Platinum Support & Care Services Ltd
2016 — 2018Earlier
Education
EDI Level 3 Diploma in Business & Administration
2014
Role
Earlier roles
Various
Up to 2016Professional memberships & certifications
AWS Certified Machine Learning — Specialty
Amazon Web Services
TensorFlow Developer
DeepLearning.AI / Coursera
PCPP1 — Certified Professional in Python
Python Institute
MBCS — Member of the BCS
British Computer Society
IET Member
Institution of Engineering & Technology
Writing
Notes & essays
Working notes from real systems — what I learned, what I'd do differently, and what's worth knowing if you're building the same thing.
-
Building Production RAG Pipelines: Lessons from Real Deployments
A weekend RAG demo is one thing. A production RAG pipeline is a different animal entirely — and the cost is mostly in the parts you can’t see in a notebook. ...
Read -
From Notebooks to Pipelines: Why I Stopped Shipping Jupyter
There’s a moment in every ML project where someone says “we’ll just productionise the notebook.” It’s the same moment, every time, that the project quietly s...
Read
What People Say
Words from collaborators
Paul took our domiciliary care reporting from scattered Excel files and manual processes to a fully automated Power BI suite. What used to take the team days now takes minutes. He understood the clinical context, not just the data — that's rare.
We needed someone who could handle the full picture — data modelling, ETL pipelines, DAX, and the front-end dashboards — and explain it all to non-technical stakeholders. Paul delivered on every front.
The AAC app Paul built changed how we think about assistive technology. The AI prediction engine was fast, accurate, and most importantly, the families found it intuitive. That last part is the hardest to get right.
Paul consolidated six disconnected data sources into one live CRM dashboard our consultants actually use every day. Adoption was instant because he involved the team from day one.
What sets Paul apart is that he builds AI systems with proper engineering discipline — version control, testing, monitoring, documentation. Too many people in this space skip all of that. He doesn't.