I design and deliver production-grade AI systems.I’m an AI engineering and MLOps specialist with a proven track record of taking complex machine learning and GenAI systems from concept to scalable, secure deployment. My work sits at the intersection of software engineering, platform architecture and technical leadership.I’m particularly motivated by the hard problems: deploying AI products at scale in regulated environments, balancing velocity with governance, and turning experimental models into reliable, high-value systems.Across logistics, energy and financial services, I’ve led and delivered AI platforms and products that have generated millions of pounds in value.Selected recognition &amp; contributions- Named one of 20 “Rising Stars in Data &amp; Analytics” (2024) by Driven by Data- Invited guest lecturer on AI and MLOps at Oxford (ongoing), Warsaw (ongoing), Columbia and Royal Holloway.- Author of Machine Learning Engineering with Python (2nd edition, Packt, 2023).- “Rising Star of the Year”, British Data Awards (2022).- “Data Scientist of the Year”, Data Science Foundation (2019).

Is "I built an agent" already like saying "I built a spreadsheet"? It can be useful and drive value, but it's also so easy to do that the fact that **it's an agent** is not a differentiator. We'll also see a proliferation of them very quickly that will be hard to control, audit and manage UNLESS you plan for that now.

How do you plan for this?

1. You standardise how, where and when people can build and deploy these. This is not the same as "introducing friction for the sake of it" - focus on the platforms across the spectrum from no-code to low-code to full software engineering projects and have quality gates for deployment. Even or maybe even especially when it's very easy to do.

2. Make it impossible to have these things do anything "real" without them being registered as entities in a registry, with permissions boundaries tied to their "identity" and the roles they are inheriting. Some large companies have said "yes this means we treat them like people in terms of identity" ... in some ways yes but in many ways no. They are still apps, they are still automated, but they are flexible and can be combined dynamically to solve problems. We should embrace this but have permissions models that work with that.

3.  I can't emphasise this one enough - you build to solve **real** problems and to support **real** objectives. Can you imagine if you had a personal objective set that you had to "build 3 spreadsheets over the next quarter" - madness. Instead it should be "help solve problem X, use what tools you need to do this". And if it's spreadsheets or agents, go drive real value!

Note: This is not to dismiss AI engineering in any way. Building solutions and platforms that *are agentic* is still a big engineering task. Just like I could build an ML model in a spreadsheet before, but building a system that solved a problem at scale with an ML model within it was a different challenge.

Thoughts? 

Happy Friday!

#agents #businessvalue #engineering #spreadsheets

Andy McMahon 

AI Engineering & MLOps Leader | Building Production AI Systems at Scale | Author of Machine Learning Engineering with Python | Oxford Guest Lecturer

taplio

Andy McMahon's Linkedin Analytics

Andy McMahon

Performance ScoreAndy McMahon's LinkedIn Performance

Andy McMahon's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Andy McMahon's Linkedin Analytics

Andy McMahon

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Andy McMahon's LinkedIn Performance

Andy McMahon's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Andy McMahon's Linkedin Analytics

Performance ScoreAndy McMahon's LinkedIn Performance

Andy McMahon's Best Posts (last 30 days)