I build the data and AI infrastructure behind modern businesses.
I help businesses take control of their data, build independent technology platforms and put AI to work. From data engineering and analytics to private AI and open-source infrastructure. Practical systems, built around your business, with no unnecessary dependencies. And one conviction underneath it all: your data is where your value lives, and it deserves to be guarded like it.
Before
Fragmented
- Data scattered across tools
- Reports that disagree
- AI stuck at the demo stage
After
One platform
warehouse Running
pipelines Running
private-ai Running
Trusted by leadership at
Where I stand
Your data is the one asset nobody can copy. Stop handing it out.
Everything your business has learned is sitting in its data: years of customers, prices and decisions that cost real money to accumulate. And the default right now is to give it away: pasted into free chatbots, piped through third-party tools, accumulating on another company's servers in another jurisdiction, under terms that change at renewal.
I think that's a mistake, and I've built my work around the alternative. Sovereign data and AI means your data lives on infrastructure you control, the models that learn from it belong to you, and anything that leaves, to a cloud service or an external API, leaves because you decided it should, with the trade-offs on the table.
Not anti-cloud. Not anti-AI. Pro-ownership.
- Who your customers are and what they pay you
- Your margins, your pricing, your pipeline
- What sells, what fails, and why
Would you email this to a stranger? Then it shouldn't be a prompt either.
From fragmented data to systems you control.
Most businesses have accumulated data across a dozen systems and a stack of subscriptions nobody chose deliberately. I build the platform underneath: four areas of work, sized to the problem, never sold as a bundle.
Data platforms & analytics
Tracking, warehouses, pipelines and reporting: one reliable picture of the business, built on data you collect and control.
Private AI & automation
AI assistants, knowledge search and LLM integrations working on your own data, with a deliberate choice about what runs privately and what uses an external API.
Infrastructure & ownership
Open-source and self-hosted infrastructure where it makes sense, with identity, security, monitoring and backups handled properly.
Machine learning & prediction
Lead scoring, lifetime value, forecasting and segmentation: models wired into daily operations, not left in a notebook.
“He was super hands-on, got our team going with several quick wins”
“Delivering remarkable results even on a shoestring budget”
“Always made an effort to understand our business”
Own the data, and the AI gets interesting.
The platform is a means, not the goal. Off-the-shelf tools give you the features they ship for everyone; once the data layer is yours, we can build the models your business actually needs, and they stay yours. For example:
Lead scoring
Score every enquiry against your own history of who actually bought, not a black-box number from a CRM add-on. Your sales team calls the right people first.
Personalisation
Recommendations and offers driven by what your customers actually do, served by your own models rather than the same generic rules every competitor runs.
Value-based bidding
Feed predicted customer value back into your ad platforms so you bid for profit, not clicks. The margin data behind the bids stays inside your platform.
Churn prediction
Spot the customers about to leave while there's still time to do something about it, trained on your retention history rather than an industry average.
Forecasting
Demand, stock, and revenue forecasts built on every signal you own, updated as the data lands. The spreadsheet guesswork retires.
The one specific to you
The model built around how your business actually makes money. No vendor can ship it as a feature, because only you have the data. That's the point.
Meet Mitch
I'm Mitch Eccles: PhD in Computer Science, building production data, analytics and AI systems for businesses since 2017. Behavioural tracking, warehouses and pipelines, machine learning, marketing attribution, and lately private AI and the infrastructure it runs on.
I'm an engineer, not a strategist with a diagram. I design the architecture, write the code, and keep the systems running in production. The technology serves the commercial problem: a system earns its keep by improving how the business operates, decides, or makes money.
I work with established B2B and B2C businesses through Pragmatik, my consultancy: companies that run on data without being tech companies. Plain English, no jargon, business first.
This isn't for everyone.
Not every business needs this, and I'd rather say so before the first invoice than after. Be honest with yourself:
Probably not a fit if…
- You want a strategy deck about AI, not working systems
- You need a large delivery team tomorrow: I'm one senior engineer, not an agency bench
- Your current tools work and the costs are under control. Honestly? Keep them
- Nobody in the business will own the platform once it is built
Probably a fit if…
- Your data is spread across systems and nobody fully trusts the numbers
- You're exploring AI but lack the data foundations to implement it properly
- You're paying for tools you don't control and couldn't easily leave
- You want senior data and AI engineering without building a large internal team
What clients say
"Aisle 3 has benefited tremendously from Mitch's expertise and dedication. Since being involved with us, he's been a driving force in solving the data challenges of online shopping aggregation, delivering remarkable results even on a shoestring budget."
"We found Mitch online and got him in to help drive our data analytics approach. He was super hands-on, got our team going with several quick wins and set a strong platform for us to grow our in house data analytics ops. I'd recommend Mitch to any business loking to build out their data infrastructure."
"I've had the pleasure of collaborating with Mitch on numerous projects over the past several years. He's partnering closely with us at Marketcheck to integrate our marketing activities with our on-site analytics and CRM, enabling us to optimize campaigns for high-value leads more effectively."
"Mitch's expertise in the AI and data space was instrumental in helping Pagnea secure grant funding, which allowed us to expand our team and take Mitch's ideas into development."
"Very smart with a PhD in copy/paste or some other clever data tech. Always made an effort to understand our business and contributed heavily towards the creation of our AI team."
Your business could be next.
Thirty minutes, no prep, no obligation. Bring a problem.
Book an Intro CallThe Sovereign Stack
The journal: practical engineering for businesses that want control over their data and AI.
Getting Started with Data Infrastructure
A practical guide to building a solid data foundation for your business, from tracking to prediction.
Take control of your data.
Thirty minutes about your business, its data and its systems. No prep, no deck, no obligation. Just a straight conversation.
Questions, answered
What do you actually build?
Data and AI platforms, end to end: behavioural tracking and ingestion to collect the data, a warehouse to store it, transformation to make it trustworthy, reporting your team actually uses, and machine learning and private AI working on top. Mostly proven open-source tools, running on infrastructure you control, so every piece is swappable and nothing holds you to ransom.
What do we get out of it?
Outcomes, not dashboards for their own sake. Lead scoring trained on who actually bought from you. Personalisation driven by your customers' behaviour. Value-based bidding that feeds predicted profit back into your ad platforms. Churn prediction, forecasting, and the model specific to how your business makes money. None of it limited to an off-the-shelf feature list.
We're not a tech company. Who runs it after you leave?
Your people, and that's the point. I build the platform to be boring to operate, document it in plain English, and train whoever will own it. And I don't vanish: I can take responsibility for ongoing operation for as long as you want me to.
Everyone pastes everything into ChatGPT now. Why swim against that?
Because of what's in the paste. Customer lists, contracts, pricing, financials: when that goes into a free chatbot it lands on another company's servers, usually in another jurisdiction, under terms you didn't negotiate and laws you don't vote on. You wouldn't hand your customer book and your margins to a stranger because the stranger was convenient. Private AI gives your team the same capability on infrastructure you control, so convenience stops costing you the asset.
When do private AI models make sense, and when do external APIs?
It's a trade-off, not a religion. External APIs are excellent for low-stakes work and fast prototypes. Private models earn their place when the data is commercially sensitive, when you need behaviour you can control and version, or when per-call pricing stops making sense at your volume. Part of my job is making that call case by case, in plain English, with the costs on the table.
What does it cost?
It depends on scope, and I'd rather give you a straight answer on the call than publish a rate-card maze. Some builds pay for themselves in retired subscriptions, others are new capability with a new cost. Either way you'll leave the intro call knowing what an engagement would look like and roughly what it would cost.
What's Pragmatik, and what's mitch.co.uk?
Pragmatik is the consultancy I founded and deliver work through: contracts, invoices and delivery run through it. This site is me: the writing, the thinking, and the track record. The engineer is the same either way.
Do you actually build things, or just advise?
I build. I've spent my career implementing data infrastructure, machine learning, and AI systems, and I keep them running in production afterwards. The advice comes free with the keyboard.