AI Solutions
Applied machine learning built around a real problem, not a demo.
Learn moreHello, I’m Humza Imran
I design and build AI systems and production software, and I train the engineers who maintain them.
Humza Imran
AI & Software Solutions Expert
What I do
Research-grade thinking, production engineering, and training that outlasts the engagement.
Applied machine learning built around a real problem, not a demo.
Learn moreProduction web platforms and APIs written to be maintained by someone else — tested, documented and handed over properly.
Learn moreWorkshops, courses and lectures that leave your team able to build without me — hands-on, with your own data wherever possible.
Learn moreSelected work
A few projects where the interesting part was not the model.
AI Solutions · 2026
A triage support tool that ranks incoming cases by urgency and, crucially, abstains when it is not confident enough to be useful.
Read the case studyAI Solutions · 2026
Retrieval and extraction across twenty years of unstructured engineering reports, with citations back to the source page.
Read the case studyAI Solutions · 2025
Surface defect detection running on the line itself, trained from a few hundred labelled examples rather than a few hundred thousand.
Read the case studyResearch
The questions I keep coming back to, in papers and in production.
Exploring Artificial Intelligence methods for developing adaptive systems and data-driven solutions to complex real-world challenges.
Exploring ML techniques to develop intelligent systems, focusing on prediction, pattern recognition, and data-driven solutions.
Reproducible training, drift detection and the engineering practice that keeps a deployed model honest over time.
Retrieval, summarisation and extraction over domain documents, with a particular interest in evaluating what these systems get wrong.
Learn
Self-paced, with quizzes and a certificate that anyone can verify.
Artificial Intelligence · Beginner
Build AI systems around a decision, not a demo.
Writing
Notes on the parts of this field that are less glamorous than they should be.
18 Sep 2026
The model is rarely the bottleneckMost projects fail on evaluation, not architecture.Define the decision firstBuild the test set before the model
Read the article17 Sep 2026
The model is perhaps a tenth of what you are shipping. The other nine tenths decide whether anyone keeps using it.
Read the article10 Sep 2026
A held-out set is only held out if it was separated before you started making decisions. Usually it was not.
Read the articleTell me what you are trying to build or teach, and I will tell you honestly whether I am the right person for it.
Get in touch