Gareth BeallLead AI/ML EngineerI build production AI systems for work where evidence matters. My advantage is that I understand both sides: fifteen years of clinical pharmacy, including intensive care, and six years of AI/ML engineering. I've been building software since I was twelve.Email GarethLinkedInGitHubDownload résumé01Clinical AIBuilt IRIS from idea to governed deliveryA clinical RAG system for Queensland Health. I took it from first prototype to governed release, and tuned every pipeline stage until clinicians could rely on it.02Rare contextFifteen years inside clinical practiceClinical pharmacy experience, including intensive care, shapes how I reason about evidence, ambiguity, human factors, and the consequences of confident errors.03Product instinctShips tools, not demonstrationsBuilt and ran clinical calculators, consumer apps, photo restoration, and small AI experiments. All of them shipped to real users.Selected systemIRISClinical AI assistant for Queensland HealthIRIS is an AI assistant Gareth conceived and built for clinicians at Queensland Health. Staff ask a question in plain language and IRIS answers from the health service's own approved documents. Every answer shows its exact sources, and when the evidence isn't there, IRIS says so.What it doesAnswers clinical questions from approved local documents, with the supporting passages shown beside every answer.How it worksIRIS checks every document it ingests, searches by exact keywords and by meaning, and re-ranks the best passages before it writes an answer.Runs in-houseThe AI models run on the hospital's own GPU servers, so sensitive data never leaves the network.Built to be safeTested against realistic clinical questions, including ones it should refuse. An honest "I don't know" beats confident guessing.Delivered for realSecurity review, user acceptance testing with clinicians, controlled releases, monitoring, and ongoing operations.Independent productsUseful things, shippedClinical productOpioid Conversion CalculatorA medication-safety tool that puts specialist opioid-conversion knowledge into a focused, dependable interface for health professionals.Visit project ↗Health AI ventureInky HealthPrivate, self-hosted document AI for health and government: search, structured extraction, and redaction for organisations whose data cannot leave their own network.Visit project ↗Consumer productNamelyA baby-name app that learns what you like as you swipe, instead of handing you an endless list.Visit project ↗AI productMy Restored PhotoAI photo restoration. Send a damaged photograph that matters to you, get it back repaired.Visit project ↗ContactLet's talk.Sunshine Coast, Queensland, Australiagareth.beall@gmail.com
01Clinical AIBuilt IRIS from idea to governed deliveryA clinical RAG system for Queensland Health. I took it from first prototype to governed release, and tuned every pipeline stage until clinicians could rely on it.
02Rare contextFifteen years inside clinical practiceClinical pharmacy experience, including intensive care, shapes how I reason about evidence, ambiguity, human factors, and the consequences of confident errors.
03Product instinctShips tools, not demonstrationsBuilt and ran clinical calculators, consumer apps, photo restoration, and small AI experiments. All of them shipped to real users.