🔸Unpacking Software Architecture

I help people deploy more value with less code.

Mike Thornton

Helping you become a better software engineer through coding challenges that build real applications.

I’ve built and scaled high-performing teams of software engineers, shaped engineering culture, and driven technical strategy in VC-backed startups and multi-billion-dollar enterprises. I’m known for taking on underperforming teams and turning them into confident, capable, and effective engineering organisations.

I combine deep technical expertise with a pragmatic, commercially driven approach to leadership. My MBA, along with experience as a founder and startup executive, enables me to align technology strategy with business objectives and deliver measurable commercial outcomes. A servant leader at heart, I focus on mentoring, coaching, and enabling teams to deliver software that drives growth and business success.

I bring:
- 25+ years of software engineering experience across a range of technology stacks. 
- 20+ years of software engineering leadership experience from Engineering Manager to VP Engineering/CTO.
- 10+ years of shipping AI based products (across Evolutionary Algorithms, Constraint Propagation, Search and Optimisation, Neural Networks and LLMs).
- 5+ years of experience representing software engineering at board level.
- 3+ years of leveraging LLMs for software engineering.

John Crickett

On Sunday, I used AI to build an app. In a couple of hours, it generated 23,000 lines of code with full documentation and 836 passing tests.

The code is amazing. The app is fully functional, well written, well structured, and has great test coverage.

At least that's what the agent told me after reviewing the project structure, completeness, and code quality.

So I fired it up and clicked the button. Nothing.

Turns out all tests passing and the agent grading its own work as brilliant doesn't mean it actually works. Now I'm debugging, filling gaps, and showing the agent what it missed.

AI can get you most of the way there, fast. But that last bit? That still needs a human who knows what they're doing. Let's not pretend otherwise.

I've loved AI for a long time. Not the hype cycle version. The real thing.

Which is exactly why I can see its flaws.

Something strange is happening in tech discourse. If you're not a full believer, you're a skeptic. Point out a limitation and suddenly you "don't get it." We've lost the ability to hold two thoughts at once: this is valuable and this has real problems.

That's not nuance. That's tribalism.

Here's the irony: we worship innovation. But innovation almost always comes from someone who loved something enough to see what was broken in it. Jobs wasn't anti-computing. The Wright brothers weren't anti-transport. Passion and criticism aren't opposites. They're partners.

AI is transformative. It's also improvable. Those two facts coexist just fine. The people who'll actually move it forward aren't the loudest cheerleaders. They're the ones asking uncomfortable questions.

That's how progress works.

AWS had a 13-hour outage in December. Some people are blaming AI.

They're wrong.

Kiro, Amazon's AI coding agent, decided to delete and recreate an environment. It took down Cost Explorer in China for 13 hours.

Sounds like an AI problem, it wasn't.

Kiro requests authorisation before taking action by default. But the engineer had broader permissions than expected. No second approval was required. The agent was treated like a human operator and given the same access.

Amazon called it "a user access control issue, not an AI autonomy issue."

This was the second incident in recent months. Amazon had been pushing an 80% weekly AI adoption target across engineering. Push hard for adoption without pushing equally hard to create the right processes and guardrails, and you get exactly this.

The AI didn't fail. The process around it did.

Three things every team running AI near production should be doing:
1. Scope permissions tightly. Least privilege applies to bots too.
2. Keep a human in the loop. Two-person approval for production changes, should still be two people.
3. Treat AI output like any other output. Review it. Test it. Run it through CI. Deploy with care.

taplio

The Battle

Mike Thornton

John Crickett

Followers

Posts

Engagements

Likes

Performance ScoreJohn Crickett's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Mike Thornton

John Crickett

Followers

Posts

Engagements

Likes

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);}John Crickett's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreJohn Crickett's LinkedIn Performance