Skip to content
Chad DarcyContact

Senior AI Engineer · Data Scientist

Fifteen-plus years turning hard, messy business problems into production systems — data and analytics, AI automation, full-stack delivery, and safe, governed AI inside an organisation. If it's a business process, I can probably automate it.

What I bring

01

Any process, end to end

Reverse-engineer a painful manual process, then design and ship the automated system that replaces it.

02

Production-grade, not demos

Security, governance, and compliance designed in from the start — systems a business can actually run on.

03

Fast and accurate

Fleets of AI agents work a problem in parallel, with independent verification before anything is accepted.

04

Full-stack range

Data and analytics, AI automation, full-stack delivery, and safe governed AI — one person, end to end.

What I do

Three ways to work together

From a fast win to a senior hand on the wheel — pick the altitude that fits where you are.

Entry

AI website refresh + a trained assistant

A modern rebuild of your site with a chatbot that actually knows your business and can answer your customers' real questions.

  • Modern, fast, mobile-first rebuild
  • An assistant grounded in your business
  • Answers your customers' real questions
Core

Custom AI for your business

Assistants grounded in your own knowledge, dashboards that turn messy data into clear answers, and automation for the manual work quietly eating your team's time.

  • Assistants on your own knowledge
  • Dashboards over your messy data
  • Automation of manual processes
Premium

Fractional AI engineer & advisory

For businesses that want AI introduced properly: architected, governed, and built to last, with someone senior owning it.

  • Senior ownership of your AI
  • Architected and governed to last
  • Introduced safely across the org

The difference

How I work

Most people picture AI as one assistant typing away in a single window. I work differently.

  • I break a problem into independent pieces and run many AI agents on them at once — each exploring on its own.
  • Then other agents check and challenge the results before anything is accepted.
  • A governance layer I design keeps the whole thing safe, auditable, and correct.

The effect is enterprise-grade quality at startup speed — fast because it's parallel, trustworthy because verification is built into the process, not bolted on at the end.

Fan out

agents explore in parallel

Verify

agents challenge the work

Govern

safe, auditable, correct

Ship

one verified result

Selected work

Proof of range — and ambition

A few systems I've built, described by what they did. The point isn't any single project; it's the spread.

Data science · MLExpedia Group

Large-scale forecasting & analytics

A global travel marketplace making decisions across millions of listings.

Production machine-learning and forecasting systems operating at marketplace scale, with results and forecasts presented directly to senior and executive stakeholders.

Models and forecasts leaders could actually plan around.

Governed AI · Platform

An enterprise platform for a regulated industry

An organisation that needed serious AI but couldn't let its data leave its walls.

A research and AI platform architected to run identically in the cloud and in fully isolated, data-sovereign environments — multi-agent AI, retrieval-grounded answers over private data, governed routing between hosted and self-hosted models, strict tenant isolation, and compliance-grade controls throughout.

Trustworthy AI where data sovereignty and compliance are non-negotiable.

Security · Formal methods

A secure-communications system with formally-verified guarantees

A messaging platform where users had to trust the server could never read them.

A system built so the server provably never holds users' keys — backed by machine-checkable security claims, formal verification of the protocol's guarantees, and an extensive automated test suite spanning two independent language implementations.

Security you can prove, not just promise.

Agents · Automation

An autonomous, multi-agent operations assistant

Messy, real-world inputs that someone had to read, sort, and act on by hand.

A system that ingests those inputs, extracts the actionable items, keeps a human in the loop for approval, and syncs confirmed actions out to the tools people already use — running largely on its own under guardrails.

Hours of manual triage handled automatically, and safely.

Agents · Decisioning

An autonomous, multi-agent decision system

High-stakes choices that needed to be made consistently, at volume, and explained.

A system that runs unattended, scoring options in parallel through specialised agents into a single calibrated verdict — with its own measurement and attribution framework built in.

Calibrated, explainable decisions at a scale people can't match by hand.

From system to software

// production-systemlive

A messy process, rebuilt as a system

Manual steps and scattered tools become a typed, tested pipeline — observable, governed, and safe to change. The system stops being a diagram and starts being software.

  • ingest → enrich → verify → ship
  • guardrails: typed · tested · observable
  • deploys: reproducible · reversible

Ask about Chad

Not sure if it's a fit? Ask.

A grounded AI assistant trained on Chad's background. Ask whether he can build what you need — you'll get a straight, capability-focused answer.

ask-about-chad

I'm Chad's assistant. Ask me what he can build for your business — automation, custom AI, dashboards, or how he keeps it all safe and governed.

Contact

Start a conversation

Tell me about the process you'd like to automate or the problem you're trying to solve. I read every message.