At a glance
About
I have built and run every stage of the path a legacy industrial business has to walk to become AI-native, from instrumenting the physical work to putting AI into production on the data it captures.
Most transformations die in the same place. The strategy arrives, a vendor platform arrives, and three years later there is a dashboard nobody opens. The failure is rarely technical. Data gets captured that was never designed to feed anything, and the capture layer itself never gets used, because nobody asked the people doing the work what would make it worth their time.
I came up through the products themselves. Twelve years designing industrial electronics and embedded systems, fire panels, power quality and fault recording instruments, data acquisition hardware, elevator IoT. That is why I know which twelve fields matter and which forty exist because someone asked for them in 2011. The last five years have been spent turning that into platforms, products, and AI at CRC Evans.
The path
A legacy industrial business becomes AI-native in a specific order. Most stop after step two. Here is the sequence, and what I built at each stage. Click a step for the detail.
IoT on welding, coating, and inspection equipment, capturing production parameters at the point they are generated. Before this, twelve years building industrial electronics and embedded products, which is how I know what is worth measuring and what is noise.
CRCE Connect, the internal operating layer. Asset management, QHSE, custom engineering tools and calculators, field operations reporting, and the rest of what a business actually runs on. Paper and spreadsheets replaced by capture at the point of work, once, with reporting downstream of it.
The software was never the hard part. Getting crews to use it under schedule pressure was, and that is what most digitization programmes never solve.
DATA360 is CRC Evans' flagship digital platform, and the point at which the internal data layer became something the company sells.
Every joint is tagged with a unique identifier and GPS position, connecting welding, coating, NDT, material traceability, and fitment analysis under one record. From the mill to the ditch, a single traceable history for every pipe and every joint. Analysis that traditionally took hours each day now takes minutes, and the platform doubles as the as-built compliance record the customer keeps.
Built on cloud infrastructure with long-range wireless collection from field stations, so data moves off the spread without anyone carrying a laptop to it.
Covered in World Pipelines, World Oil, Pipeline & Gas Journal, and The Australian Pipeliner.
Assisted defect recognition for radiographic and ultrasonic inspection, reading the data the earlier stages capture. Retrieval-grounded assistants answering from standards, procedures, and project records with citations back to source. Forecasting and anomaly detection on operational and financial data.
Agentic AI in production across the business: a chief-of-staff agent for asset management that orchestrates multiple specialist agents, a market intelligence agent, and an agent working over the data repository that supports customer reporting.
Built on AWS with Bedrock, Lambda, and pgvector on RDS, running inside the company's own cloud environment so proprietary and customer data never leaves it. Role-based access, auditability, and data protection are designed in from the start, not bolted on. The emphasis is on production. Pilots are easy.
Four peer-reviewed papers on machine learning for weld inspection, in collaboration with the Electrical and Computer Engineering department at Lamar University. The work runs on a real proprietary industrial dataset from live girth weld inspection, not laboratory samples, which is what most published work in this field lacks.
A granted patent in three jurisdictions. Two further conference papers in progress.
Research keeps the products honest. The products keep the research grounded in data that came off a real job.
An AI Centre of Excellence and a citizen developer programme. Audience-specific training paths, playbooks, a weekly practice cadence, and governance that does not strangle what it governs.
The goal was never a central AI team taking tickets. It was a hundred people who can build the small thing themselves.
Built outside work
Mary Care
New venture · Co-founder Founded 2026A connected care platform for families supporting someone living with dementia. Three apps - a simplified Android experience for the person being supported, Android and iOS companion apps for the family around them - plus a web portal.
Built by two engineers, live on Google Play and the App Store. A registered UK company, ICO registered. I am the co-founder and main programmer alongside the other founder.
It is here because it is checkable. Everything above happened inside one company. This is what the same approach looks like starting from nothing, this year.
Experience
Eighteen years in industrial technology. The first half building the products that generate the data. The second half building what makes it useful. Click a role for the detail.
Lead digital transformation for an industrial equipment and services business operating across EMEAA and the Americas. Own the internal operating layer, the customer-facing data platform, and the AI running on both. Work directly with engineering, QHSE, field operations, project finance, HR, and legal, and set company digital strategy with the leadership team. Founded the AI Centre of Excellence and the citizen developer programme. Team of 20+.
Built and shipped CRCE Connect and the foundations of DATA360. Instrumented welding, coating, and inspection equipment; replaced paper and spreadsheet workflows with capture at the point of work.
Ran 10+ concurrent programmes, mentored four project managers, and worked across a 15-stakeholder customer account. Directed a power quality meter programme with 10+ engineers that generated over $1M in revenue, and an IoT platform build with 25+ engineers over two years. Presented quarterly reviews to customer CTOs and presidents. Sustained 4+ out of 5 customer satisfaction across the portfolio.
Built a condition-based elevator monitoring IoT system over three years with a team of 15+, spanning sensors, wireless hardware, firmware, and iOS. Cut elevator servicing time by 50%. Co-authored the patent on elevator car location determination using RFID.
This is where the pattern started. Sit with the service engineers, learn how maintenance actually works, then decide what the system should measure.
Led 10+ engineers across industrial product programmes: high-speed mixed-signal display drivers for fire panels, UL and ULC recertification, and environmental compliance across 4,000+ SKUs for a large conglomerate. Built automated reporting that saved the customer $400K+.
Lead designer on a low-noise data acquisition system with adaptive noise cancellation, FPGA, and multichannel ADC and DAC. Earlier, carrier boards and power quality and fault recording products. Requirements through schematics, PCB, board bring-up, and EMI/EMC qualification.
Skills and stack
Research and patents
Applied machine learning for industrial inspection, and the digital transformation work behind it. Four peer-reviewed papers on a real proprietary industrial dataset, a patent granted in three jurisdictions, two conference papers in progress.
Three papers on one dataset, three years, each building on the last.
RFID tags at each hoistway landing paired with a car-mounted reader, giving a low-cost retrofittable position reference that triggers and locates condition-monitoring data by floor. The same idea underneath all the work since: attach operational data to a location so decisions can be made per asset instead of in aggregate.
Education and recognition
Also: PGP in Artificial Intelligence, Texas McCombs School of Business (2022). PGP in Management, IMT Ghaziabad (2021).
Project Management Professional (PMP)
Get in touch
Roles, advisory, research collaboration, or a straight technical question. Everything sent here reaches me directly.