AI Built on Engineering, Not Hype
I build AI systems grounded in real research and practical engineering. My published work includes MFCC-based voice recognition for industrial environments and algorithm optimisation for robotic assembly — and I've shipped tools like TranscribAIr, which uses Whisper and LLMs to automatically categorise educator feedback.
What I Build
- Speech & Audio Processing - Transcription, voice recognition, and audio analysis systems using Whisper and custom pipelines
- Computer Vision - Image analysis, photogrammetry, and visual inspection systems for research and manufacturing
- LLM Applications - Intelligent tools that use large language models for classification, extraction, summarisation, and generation
- AI-Powered Automation - Systems that replace manual processes with intelligent decision-making
- API Integrations - Connecting your applications with OpenAI, Anthropic, and open-source models
Cloud or On-Premises
Not every AI project can send data to the cloud. I work with both:
- Cloud APIs: OpenAI, Anthropic Claude, Google — when speed and capability matter most
- Local deployment: Ollama, Llama, and other open-source models — when privacy, cost, or compliance require on-premises processing
My Background in AI
This isn't a side interest bolted onto web development. My AI work comes from:
- Published research in voice recognition reliability for industrial human-robot interaction
- Computer vision experience from years of photogrammetry and image analysis at Swansea University
- Machine learning applications in algorithm optimisation for robotic manufacturing
- Active development of open-source AI tools like TranscribAIr