Richard J. Doherty
I've carried accountability for mission-critical systems at global scale — and today I'm building and shipping AI-assisted products solo. VP/CTO-level delivery, paired with current, hands-on data science and applied AI. Few people hold both ends of that spectrum.
For twenty-five years I led technology for some of the world's largest enterprises — VP-level ERP design and deployment at GlaxoSmithKline, global BI consolidation across six countries at Diageo, and CTO of a venture-backed SAP marketplace. I learned how mission-critical systems are governed, delivered, and trusted at scale.
Then I made a deliberate pivot. I returned to study for a Master of Data Science at RMIT, applied machine learning to clinical and operational data in regulated medtech, and spent eight-plus years teaching SQL and database systems at university level — earning a Dean's commendation and consistently top student ratings.
Today I work at the frontier of applied AI: advising organisations as a fractional CIO, and building and shipping production software solo under my own company, Kyvara — using modern AI tooling with the engineering discipline I carried out of the enterprise world. The combination is the point: I can be accountable for the outcome and write the code that gets there.
How I can help
Each lane is backed by verified delivery — proof, not adjectives. Pick the one that matches your need.
Turn data and large language models into decisions and shipped product — scoped honestly, evaluated rigorously.
The proof: Master of Data Science (RMIT). Applied ML on clinical data in regulated medtech at IMRA Surgical. Five LLM document generators built and evaluation-gated.
Senior technology leadership on demand — strategy, governance, and delivery without a permanent hire.
The proof: VP of ERP Design & Deployment at GSK (48 consultants, £5M+ budget). CTO of a $40M-funded startup. Advising clients across AU, UK & USA.
Large, regulated ERP programmes taken end to end — design, release governance, and multi-country rollout.
The proof: 25+ years full lifecycle. Global BW/BI consolidation across 6 countries at Diageo. Release management across 5 regulated ERP systems.
Bring stakeholders, students, and delivery teams along — translating complex technology into outcomes people trust.
The proof: 8+ years teaching SQL & databases at university level. 4.97/5 "effective teacher". Dean's-commended 4.95 average. Capstones for Home Depot, FedEx & Regions.
Track record
A scannable history of verified roles, leading with impact. The oldest roles are condensed below.
Fractional CIO and AI advisory for clients across Australia, the United Kingdom and the United States.
ACN 697 072 049. Shipped TenantEvidence (iOS) and Kirevra Press (web + YouTube) — roughly seven weeks from incorporation to App Store submission, solo.
Applied machine learning on clinical and operational data in regulated medtech (robotic surgery).
Executive manager and subject-matter expert for IT across a multidisciplinary research programme.
Taught Database Systems: SQL, MySQL, SQL Server, ERD and Power BI. Student evaluations 5.8–6.0 / 6.0.
Database, SQL and Agile. Dean's commendation (4.95 average); 4.97/5 "effective teacher". Client capstones: Home Depot, FedEx, Regions, Stern Agee.
Recent work · building now
The freshest, most differentiating material: two products shipped solo, and the disciplined AI-assisted method behind them.
The method — the real differentiator
Both products were built solo using a disciplined, multi-session AI-assisted workflow — with me as the deciding human in the loop, and verification against real ground truth at every gate.
Separate planning, coding (Claude Code) and design/deploy sessions — each in its own scope, with me as the courier between them. No AI surface acts without my ratification.
I own strategy, architecture and the specification layer; direct code generation; then review and test every change before it merges. Authorship stays with the person directing and verifying.
Every "fixed" claim is checked on-device or in production — not just in the diff. This caught defects neither the code nor the review found alone.
Contemporaneous records per session and a running ledger of defects found and closed — an auditable trail, not just a finished artefact.
This is applied AI-assisted development with engineering discipline — not ML model engineering, and not the SQL/relational work that sits in my data-science and university-teaching background. Both are real; I keep them clearly distinct.
Proof & credentials
Every figure here is verified and public-safe. No invented metrics.
Skills & technology
Contact
Whether you're hiring, raising, or looking for an advisor — the fastest route to a conversation is a direct message.
UK & US numbers available on request.