I build AI systems that hold up — and teach teams to run them.

What I do

What I actually do

1. 01

   Build AI that works in production

   Through Axys Analytics I design and deliver the system around the model: agents, retrieval, evaluation, routing, guardrails and the operating model. It’s built for the messy middle of a real business, not for a demo.

2. 02

   Turn your archive into answers

   Most firms already have the expertise. I structure it into a knowledge base, then turn it into answer-first SEO/AEO content that ships across web, social and media. People still approve whatever goes live.

3. 03

   Teach teams to use AI properly

   Since 2022 I’ve taught practical LLM work in public at PromptEngineering.org, through writing and open libraries like Partials, Agents and Miniscripts, for 20k+ subscribers. AgentEngineering.org does the same for agent systems.

4. 04

   Keep it accountable

   I came up through IS audit (CISA, top 10%) and internal-audit leadership, with an MBA on top. Governance, controls and evidence go in from day one instead of getting bolted on later.

Proof

The work behind it

• PromptEngineering.orgPractical LLM teaching in public, with writing and open libraries.See it on Work →
• AgentEngineering.orgAgent systems: specialists, memory vs. knowledge, and the harness around the LLM.See it on Work →
• Axys AnalyticsProduction AI for companies, covering evaluation, routing, guardrails and the operating model.See it on Work →
• A 30-year Caribbean firmFrom archive to knowledge base to answer-first content across web, social and media.Read the case →

Skills

What I’m good at

The longer list, with somewhere to check each one.

• 01

  Prompt engineering

  I treat prompts as systems: templates, examples, versioning and tests, not one clever line in a chat box.

  • System prompts
  • Few-shot design
  • Prompt templates
  • Context engineering
  • PseudoLangs

  Open libraries Partials, Agents and Miniscripts, plus the Prompt Engineering Masterclass.

• 02

  AI engineering & agents

  I design agent systems that split work across specialists, keep memory apart from knowledge, and check their output before anyone relies on it.

  • Multi-agent orchestration
  • Agent memory
  • Human-in-the-loop design
  • Routing

  GAINs and HCIN on real agents · The Decision Layer · AgentEngineering.org

• 03

  Evals & AI quality

  I test AI the way you’d test anything people depend on: read the failures, write the checks, and gate every release on them.

  • Error analysis
  • Offline & online evals
  • Regression gates
  • Tracing

  Error analysis: the first eval · Evaluating trajectories · Regression testing for agents

• 04

  RAG & retrieval

  I build retrieval that gives the model the right source at the right moment, and shows that source with every answer.

  • Chunking & indexing
  • Graph RAG
  • Cited answers

  A 6-stage RAG pipeline · A brief guide to RAG · Axys: Ask your business

• 05

  Workflow automation

  I turn repeat work into AI workflows that draft, process, reconcile and route, with a person approving anything that matters.

  • Agentic workflows
  • Make & Zapier
  • Approval gates

  The ACE framework for durable AI workflows · Axys: Put your business to work · co-founded LifeBot Automation (2018–19)

• 06

  Private & on-prem AI

  I deploy AI inside a private cloud or on a client’s own servers when the data can’t leave the building.

  • Data sovereignty
  • Role-based access
  • Audit logging

  Cloud or on-premise AI? · Axys security model

• 07

  Solutions architecture & client delivery

  I sit with the client, map how work really happens, then design and ship the system myself, the way a forward-deployed engineer would.

  • Discovery interviews
  • Build blueprints
  • Fixed-price scoping

  Founder / AI Solutions Architect, Axys Analytics since 2018 · Law-firm AI build

• 08

  AI strategy & delivery

  I help leaders pick the few AI use cases worth doing, then get them built, adopted and measured.

  • Readiness assessments
  • Use-case scoring
  • 90-day roadmaps
  • Vendor & demo reviews

  Axys readiness assessment: a 7-dimension scorecard, use-case shortlist and 90-day roadmap · client work across insurance, finance, retail, automotive and oil and gas

• 09

  AI adoption & change

  AI is mostly people and process, so I start with the frontline, train internal champions, and keep score every month after go-live.

  • Frontline-first rollout
  • Champion programs
  • Operating model

  Axys frontline-first method · Monthly AI operating rhythm

• 10

  Knowledge bases & AI search

  I turn a firm’s documents into a knowledge base, then into answer-first content that search and answer engines can cite (SEO/AEO).

  • SEO/AEO
  • Technical SEO & schema
  • Publish once, use everywhere

  The 30-year Caribbean firm case · The new SEO: owning the answer

• 11

  AI governance & risk

  Evaluation, guardrails, access controls and audit trails, from someone who did IS audit for a living.

  • NIST AI RMF
  • ISO/IEC 42001
  • Internal audit & controls

  CISA (top 10%) · former Lead IS Auditor · Axys governance, aligned to NIST AI RMF and ISO/IEC 42001

• 12

  AI security

  Prompt injection has no patch, so I design for containment: limited permissions, checked outputs and no blind trust in tool text.

  • Prompt injection
  • OWASP LLM & Agentic Top 10
  • Least privilege

  Agent security and the OWASP Agentic Top 10 · cybersecurity and IS audit background

• 13

  AI education & training

  I teach practical AI to people who have to use it at work: courses, playbooks and workshops, for beginners through to executives.

  • Curriculum design
  • AI literacy for leaders
  • Playbooks

  20k+ subscribers · AI Foundations course · Prompt Engineering Masterclass

• 14

  Audience & community building

  I grew PromptEngineering.org from a domain registered in March 2022 to 20k+ subscribers, by publishing useful work in the open.

  • Newsletters
  • Open libraries
  • Developer education

  PromptEngineering.org · The Prompt newsletter

• 15

  Content strategy & brand voice

  I build the voice guides, research sources and templates that let a team publish with AI and still sound like themselves.

  • Tone of voice
  • Trend research
  • Editorial systems

  Tone of voice framework · AI trend forecasting guide

• 16

  Digital marketing & analytics

  Twenty years of SEO, content and analytics work before AI, which is why I measure what ships.

  HubSpot Content Marketing and Semrush certifications