About
John Carney
Independent Principal AI Engineer · Founder, PDFTA · Manchester, UK
John's value is pattern recognition accumulated across dozens of organisations and sectors — applied at the organisational level. He's seen what works and what breaks. No internal hire can bring what that combination offers.
£83.1m
Incremental revenue from a single recommendation system
20+
AI/ML engagements across eCommerce, FinTech, HealthTech, and ClimateTech
4 weeks
To turn a 15-month stalled project into a working system the team could own
Weeks
To catch a compliance blocker before months of investment were committed
Background
From plant genetics to AI/ML consultancy.
John completed a PhD in Plant Genetics at the University of Reading (2011–2015), where he built statistical modelling and ML skills through research. His career moved from scientific computing into commercial data science.
He founded PDFTA in April 2018. Since then, he has worked across eCommerce, ClimateTech, FinTech, and HealthTech — accumulating the cross-sector pattern recognition that sits at the core of PDFTA's value.
The referral script is straightforward: John works with organisations that want to realise real value from AI — not just ship one project, but build the capability to keep delivering on it. He helps them put the teams, tooling, and processes in place so that the AI work compounds over time rather than staying isolated.
Community
Global presence in the data community.
Chair — PyData Strategic Committee
Jul 2025245+ global chapters; John led its creation
Conference Chair — PyData London
Ongoing650+ attendees; selects programme from 300+ submitted talks
Co-Founder — PyDataMCR
Feb 2018Manchester's data science community
Co-Founder — Field of Play
Oct 2024Sports analytics community; 300+ attendees at Bridgewater Hall
What clients say
Outcomes that speak for themselves.
"John's recommendations didn't just offer temporary fixes; they fundamentally transformed our organisation. His work has become the essential foundation for our current solution, significantly improving data processing speed, accuracy, and overall system maintainability."
"Four weeks with John would have saved 12 months of frustration and loss."
Case studies
Pattern recognition across sectors.
Social Impact
Supply chain ethics
Rescued a 15-month stalled project in 4 weeks by right-sizing the architecture to the team that had to own and maintain it.
The existing system had become too complex for the team to operate. A simpler architecture, built around what the team could actually own, unlocked what 15 months of effort couldn't.
HealthTech
Neurodiversity
Delivered a serverless S3/Athena data warehouse on a fixed-price basis, unblocking ML model training and enabling MVP launch.
A startup needed a data foundation before they could build models. A lean, serverless architecture gave them what they needed without the operational overhead of a managed warehouse.
FinTech
Insurance
Identified a reinsurer compliance blocker within weeks, preventing significant wasted investment in a path that could never clear regulatory approval.
Pattern recognition across sectors. A constraint visible to someone who'd seen similar regulatory dynamics elsewhere — caught before months of work were spent on it.
Public talks
Sharing the work publicly.
Delivering Valuable ML Products
PyDataFrom Inception to Production
PyDataMachine Learning in Production
PyDataDeterministic Products in a Probabilistic World
Leeds Data Science, May 2026