A working history of applied AI.

Facilitated a mapping workshop for the International Atomic Energy Authority in Vienna. The goal was to organise a large set of documents. We categorised each document to see where it fit into an overall framework. This allowed the document set to be rationalised, performing such tasks as eliminating duplicates and contradictions. The Mapping process also identified any gaps - finding any parts of the top-level framework that did not have matching documents at the detail level.

Mythforge Studios is an AI Elevate game-development initiative that created Shattered Arcana, an ambitious Unreal Engine 5 turn-based fantasy 4X strategy game concept spanning multiple planes of existence, dozens of races, dragons, magic schools, world systems, diplomacy, trade, siege warfare, weather, and procedural world generation.

The project produced a full pre-production package including the game design document, world bible, art bible, sound direction, production roadmap, sprint plans, QA gates, UE5 architecture, C++ module scaffolding, gameplay-system prototypes, automation tests, AI-assisted art pipelines, generated concept assets, 3D model workflows, and a themed public-facing studio/game website.

We delivered a practical cybersecurity workshop designed for non-technical employees — the people most likely to be the entry point for an attack.

For Starbucks, what set this engagement apart was a clear focus on the real-world, human factors that drive most security breaches — specifically: understanding why 95% of incidents stem from human error; how attackers manipulate trust and attention; and which practical habits truly make a difference. The jargon-free workshop was structured around tangible scenarios such as spotting phishing attempts, recognizing social engineering tricks, identifying modern malware, maintaining strong password hygiene, using multi-factor authentication, and responding effectively to incidents.

We designed and delivered a bespoke one-day Microsoft 365 Copilot programme for the Northern Ireland Audit Office, built entirely around public audit workflows rather than generic productivity scenarios.

Participants worked through audit-specific use cases, including financial statement comparison, risk extraction, performance audit report drafting, and responsible AI governance in a public-sector context. They finished the day building custom Copilot agents grounded in their own organisational knowledge.

The AI Advantage series helped participants build practical confidence with AI across a sequence of focused seminars: AI in Plain English, use-case labs and prompting foundations, building AI copilots, handling hallucinations, data safety and ethics, budget-friendly AI tooling, proposal and report acceleration, and rolling out AI successfully.

We designed and delivered a hands-on debugging course for Salesforce engineers focused on production AI systems, built around a purpose-built multi-service RAG application that served as a live lab environment.

Rather than teaching from slides, participants instrumented real microservices with OpenTelemetry, diagnosed pre-broken failure scenarios across vector databases, LLM proxies, and orchestration layers, and used Langfuse and Grafana to trace root causes in under two minutes.

We designed and delivered a hands-on technical workshop for Salesforce engineers tackling one of the hardest problems in modern software: shipping LLM features with confidence.

The workshop moved engineers from “vibes-based” development to systematic quality assurance — covering metric-based unit testing with DeepEval, golden dataset regression suites, Langfuse observability, and structured output validation. Feedback included: “I appreciate the teaching style” and “you certainly stand out as a wonderful instructor!”

We delivered a Cybersecurity Fundamentals workshop for the Engineering Equipment and Materials Users Association, designed to make cyber risk practical and understandable for non-technical and technical colleagues alike. The training focused on the human factors behind most breaches, using real-world scenarios to build confidence in spotting phishing and social engineering, strengthening password and MFA habits, and responding quickly to potential incidents. The result was a more security-aware workforce and a stronger day-to-day security posture to support EEMUA’s wider industrial cyber security work with its member community.

GigForge is an AI‑powered development agency that builds custom software using a coordinated team of autonomous AI agents, with human oversight at every approval gate to keep quality and direction on track.

Clients describe what they need in plain language, and GigForge’s agent team handles the full lifecycle—from requirements and design through coding, testing, Dockerisation, and deployment—across full‑stack web apps, SaaS platforms, AI/ML integrations, DevOps/CI/CD, RAG chatbots, and bespoke business agents.

Delivered a Mastering LLMs training engagement for a private client.

Delivered an AI-assisted software development programme for European space-sector teams, helping technical participants apply modern AI development workflows to demanding scientific and engineering contexts.

We designed and delivered a bespoke three-part generative AI workshop series for Salesforce, spanning strategic AI innovation, practical AI application development, and AI-assisted software engineering.

The series covered business use-case identification, ROI and pilot planning, AI APIs, embeddings, RAG, production considerations, coding assistants, agentic workflows, and emerging practices such as vibe coding. It gave participants a shared language for AI adoption, practical tools for evaluating opportunities, and confidence to move from experimentation toward real business and engineering impact.

Course Creator is an enterprise learning platform with AI-powered course generation, isolated Docker-based lab environments, multi-tenant RBAC, analytics, RAG, and local LLM inference.

The platform uses a 16-service microservices architecture with Python FastAPI services, PostgreSQL, Redis, React/Vite, Nginx, Docker Compose, Claude/OpenAI integration, local Llama 3.1 inference via Ollama, a RAG pipeline, knowledge graph services, AI assistant capability, and multi-format export including PowerPoint, PDF, Excel, SCORM, and ZIP. It has 400+ automated tests across services.

Delivered a Mastering LLMs course for Multiverse Computing.

Delivered a Mastering LLMs course for EY.

Delivered an AI-assisted software development course for European Space Agency participants.

Delivered a Mastering LLMs training engagement for a private client.

NIMCP (Neuro‑Inspired Modular Control Protocol) is an open‑source “artificial brain” architecture that trains six heterogeneous neural network types—spiking, liquid, convolutional, Fourier, Hamiltonian, and adaptive—together as a single 2.5‑million‑neuron system on a single desktop‑class GPU. It uses biologically inspired plasticity to enable continual learning without catastrophic forgetting, organizes training through a four‑stage developmental curriculum that mirrors human cognitive development, and embeds a non‑removable ethics and governance stack so that safety is guaranteed structurally rather than by behavioural alignment alone.

The system integrates rich sensorimotor capabilities, including 12 sensor types, ROS 2 robotics bridge, swarm and federated operation, and a brain‑native language layer. It is documented by technical papers covering its mathematical foundations, multi‑network training methodology, emergent spiking dynamics, safety architecture, and socioeconomic impact, all designed so that university labs and independent researchers can reproduce and extend the work using modest compute resources.

We delivered a custom three-day training to help instrument scientists, senior developers and software engineers build with AI. The course was tailored to space science, utilised astronomical datasets, and helped show how the scientists present could meet their data and development challenges with modern AI development tools.

Participant feedback included: “Excellent course: the content was very interesting and immediately applicable to daily work. Additionally, the instructor was outstanding, explaining everything very clearly and showing genuine interest in our needs and progress throughout. Truly satisfied. Thank you!”

We authored the Generative AI curriculum for Morgan Stanley’s prestigious Technology Analyst Program. This introduced developers to GenAI, prompt engineering and AI-assisted software development, and is delivered across the globe.

AI Elevate delivered a tailored workshop for the European Space Agency on advanced AI algorithms for image analysis, designed specifically around space telescope and satellite imagery use cases rather than generic machine-learning theory.

The session moved from a practical CNN refresher into deeper techniques including variational autoencoders, GANs, attention mechanisms, transformers, and U-Net segmentation, with examples such as generating galaxy shapes from redshift images, captioning satellite imagery, enhancing telescope images, and identifying regions of interest in satellite data.

We developed and delivered a workshop for Vorboss, a business-focused ISP in London. The “Mastering Prompt Engineering” curriculum enabled users across sales, marketing, product development, customer support and other business domains to use AI to enhance productivity, improve customer service, and increase quality across their operations.

We designed and built a personalized web application for flower farming enthusiasts, with a focus on the growing “slow flowers” movement — home growers and small-scale sellers who want to compete with commercially imported flowers.

The app pulls live weather data, seasonal almanac information, and best-practice content from farming websites to generate advice tailored to each user’s specific region, available growing space, and chosen flower varieties. It helps users schedule seed starting, succession sowing, transplanting, and harvesting across the season, then alerts them to frost events and anomalous weather with actionable steps.