AI that ships.
Built by an engineer.

I'm an engineer who builds AI and automation on top of the systems you already have. The kind of project a Big Four firm quotes at six figures and nine months, I usually ship in three to six weeks for a fraction of the cost — directly, no associate layer, no roadmap theater.

Start the Conversation See the Work →
3–6 weeks Typical Delivery
~20% Of Big Four Cost
One Senior No Associate Layer
Working software, not slideware◆ Code in production, not concepts in PDF◆ Engineer-to-engineer, not associate-to-account-manager◆ Weeks, not quarters◆ Working software, not slideware◆ Code in production, not concepts in PDF◆ Engineer-to-engineer, not associate-to-account-manager◆ Weeks, not quarters◆
What I Build

Three things, done well.

No menu of buzzwords. No "transformation roadmaps." Just the work that ships actual value, in weeks, on top of the infrastructure you already run.

01

Modernize without ripping it out

Your aging system probably doesn't need replacing. It needs a layer on top. I add the modern part — APIs, dashboards, AI tools, automation — without forcing rip-and-replace pricing on a system that already works.

Modernization
02

AI tools that know your data

Custom Claude/LLM tools wired into the systems you already run — your historian, your CRM, your ticketing, your spreadsheets. Not a chatbot you re-teach every conversation. Tools that already know what you mean when you say "that asset" or "last quarter."

AI Build
03

The thing eating your team's day

Manual reports, copy-paste between systems, end-of-month spreadsheet rituals. Whatever's burning your team's hours, I build the thing that does it for them. Usually shipping in two to four weeks.

Automation
Selected Work

What this looks like in practice.

A few recent projects, anonymized. Each one shipped, runs in production, and replaced a workflow that would have cost an order of magnitude more through traditional channels.

Industrial operations · 8+ field assets

A natural-language query tool over a live operational historian.

Before: ops leadership wanted to ask questions of years of equipment data. Every question required an engineer to write SQL by hand. A simple weekly trend check ate half a day.

What I built: a Claude-powered query tool wired directly into their MQTT historian. Custom domain tools, WebSocket streaming, schema and naming conventions baked in. Ops team asks in plain English, the tool queries the live data.

Result: live in production. Half-day workflow now takes thirty seconds. The engineers who used to write the SQL are doing actual engineering again.

Industrial operations · legacy SCADA migration

Migrated 10 field units off a sunsetting platform — in weeks, not quarters.

Before: the vendor was sunsetting the legacy SCADA platform their 10-unit field deployment ran on. The vendor's preferred migration partner quoted a multi-quarter project at typical consulting rates.

What I built: a parallel migration to a modern Ignition stack. Native gateway alarm rosters replaced the tag-based legacy ones. One parameterized alarm pipeline shared across all 10 units. Theme-responsive UI. Reusable codebase for the rest of the fleet.

Result: shipped, running, no operational losses during cutover. Fraction of the quoted timeline and cost.

9-figure operations company · cross-system data

One platform asking questions across SharePoint, ticketing, and two databases.

Before: operational data scattered across SharePoint, an MSP ticket system, two databases, and several spreadsheets. No way to ask questions that crossed sources without manually exporting and joining each time.

What I built: a three-layer platform — ingest, data, query — on FastAPI. Schema-per-source for clean isolation, master keymap for cross-source joins, MCP exec layer that lets Claude tool-call directly into it.

Result: cross-system questions answered in plain English, on demand. No new SaaS subscriptions; runs on existing infrastructure.

Onyx, resting on the couch
In Memory of Onyx
The Person You'll Be Working With

Senior engineer.
Hands-on the whole way.

Most AI consultants haven't shipped real software in a decade. I never stopped.

I've held senior leadership roles — VP at an engineering consulting firm, Director of Controls Engineering at a 9-figure technology company — and kept my hands on the keyboard the entire time. The deliverables left my account, not somebody three layers below me. That's still how I work.

Onyx Automation is named after my Rottweiler, Onyx — a dog defined by unwavering loyalty, quiet strength, and fierce protectiveness of those she loved. Same temperament I bring to every engagement: I learn your business deeply, build the thing that actually solves the problem, and stay around long enough to make sure it earns its keep.

No associates. No slide decks. No nine-month "transformation roadmap." Just an engineer doing engineering for you — fast, well, and at a fraction of what the brand-name firms charge.

"For Onyx. Who was always the most loyal one in the room."
Questions

Common Questions

Don't see your question here? Reach out directly — we're always happy to talk through your specific situation.

Why are you so much cheaper than the firms I've talked to? +
Because I'm a senior engineer doing senior engineering work directly for you. The big firms charge what they charge because they need to pay three layers of people you'll never talk to — partners, managers, and a rotating cast of associates. I don't have those layers. You get my time, not a junior's.
How do you compress a nine-month project into a few weeks? +
By skipping the parts that don't ship code. No discovery deck, no governance committee, no week of stakeholder interviews to find out what you've already told me directly. I scope what's actually needed, build it, deploy it. The longest stretch on most projects is waiting on IT access.
Our system is old. Will you actually work with it? +
That's specifically what I do. Most of my work is adding modern capability — APIs, AI tools, automation, dashboards — to systems that were built before this stuff existed but still run the business. I'm not trying to sell you a replatform. If your current system works, we keep it.
Are you a one-person shop? +
Yes — and that's the feature, not the bug. You talk to the engineer building it, not an account manager translating between you and a vendor team. When something needs to change, the change happens that day, not after the next sprint review. For projects that genuinely need more hands, I bring in trusted specialists I've worked with before — and you still talk to me.
What if my project isn't really an "AI" project? +
Half of what I'm hired for isn't AI — it's automation, data plumbing, internal tools, and modernizing things that were already broken before AI was on the table. I take the work that ships value, regardless of buzzword. If AI fits, it fits. If it doesn't, I won't bolt one on for the sake of it.
Let's Talk

Ready to Transform?

Tell us a bit about your business and what you're hoping to achieve. We'll be in touch within one business day.

✦   Message received. We'll be in touch within one business day.
Prefer email? Reach us at hello@onyxautomation.ai