How AI Enabled Me to Go Back to Engineering After Four Years in Management

I recently made a career transition. Having been Head of Engineering for around four years, in two different businesses, leading entire departments of multi-disciplined engineers, I recently started a new role which is hands-on, building a new platform from scratch as a team of one.

Reflecting on this transition and the last few years in leadership, I feel like the advent of AI coding agents has been a massive enabler in me making the switch back. Here I offer some insights and advice from my journey in the hopes that it inspires others to do the same, and also offer my thoughts on the future of technical leadership.

The managers trap

Charity Majors wrote an article a few years ago about the Engineer/Manager Pendulum - the idea that promotion to management doesn’t have to be a one way street, engineers ‘promoted’ to management should switch back to engineering in the future and vice-versa. I always loved the notion, but never really saw a way to make that transition myself.

I’m willing to bet that a lot of senior leaders have experienced the same feelings I did. Yearning for the days when I was ‘just an engineer’ and didn’t have to deal with people and meetings. Dreading the end-of-year reviews and performance cycles, knowing that people would never be happy with the outcome. But I accepted that it was part of career progression, that I enjoy the influence, I definitely enjoy the paycheck and the company cars. I felt like my time as an engineer was over and I was stuck as a manager for life.

I don’t hate management. Being able to genuinely help people grow, enabling engineers to progress to more senior levels, there’s a real satisfaction in that. I love the influence and being able to make architecture and strategy decisions and guide the development of whole systems, as opposed to being at the whim of someone else's decisions.

But at the same time I always wanted to get back to engineering. I’m pretty introverted so I’d get to the end of most weeks feeling emotionally drained. Not because I hated people, but because management just doesn’t give the same dopamine hit that engineering does. You don’t get those fast feedback cycles and tangible outputs from your work.

As much as I loved the idea of building and keeping my skills sharp, the reality is that family and home life takes priority and there’s little time and energy left to code.

For most people it ends there. They accept their fate as a manager and gradually get further and further from the coalface.

AI changed the equation

I did try to keep my skills sharp. I lost count of all the times I’d try and build something in a new framework or smash through a tutorial. I must have tried to learn Go about a hundred times - I got somewhere, but it always felt so overwhelming. I’ve spent a fortune on Udemy courses that sit half-finished in my account somewhere.

Once you’ve been hands-off for more than a few months it basically feels impossible to get back there. The mere thought of having to do a tech test in an interview was enough to bring me out in a cold sweat.

But the advent of AI coding agents changed things.

Building with tools like Claude Code enabled me to accelerate learning by doing. I could build, then get an explanation, then build some more. It gave me genuine confidence.

It also reframes what engineering looks like. I don’t need to memorise specific syntax because who the hell does that anymore - you’ve basically got the worlds most knowledgeable pair programming partner at your fingertips.

Making the switch

I’ll be the first to admit that luck was on my side when it came to transitioning to my new role. Firstly my hand was somewhat forced - in my previous role I was offered a settlement agreement to move on after a PE group got involved (my thoughts on that in a future article!). So I was forced into the job market to some extent, albeit with a healthy runway.

I also had an awesome recruiter on my side who was able to tell my story and advocate for me with hiring managers.

Finally the interview process didn’t involve a tech test. It was more about talking through my experience and a technical conversation, which suited me far better.

All of which combined to see me land in a brilliant new role, in a domain I was keen on (fintech), with hands-on engineering and being a team of one.

Why experience matters more than ever with AI

There's a narrative on LinkedIn and Twitter (yes I still call it Twitter) that AI levels the playing field so much that experience doesn't matter anymore. That anyone can build anything now. I think that's dangerously wrong.

AI is incredibly effective when you know the right questions to ask. When you've seen how entire systems fit together. When you can look at what it's produced and immediately spot what's missing, what won't scale, what's going to cause problems six months down the line. That ability comes from years of experience, in both engineering and leadership, and no amount of prompting skill replaces it.

I have always been a generalist as an engineer. My time in management actually made this even more pronounced. As Head of Engineering I'd led teams across development, testing, DevOps/infra, security, UX, and data. I wasn't deep in any single stack, but I had a breadth of understanding across all of them. I'd seen how architectural decisions played out over time. I'd been in the room when systems failed and when they succeeded, and I understood why.

That breadth turns out to be exactly what AI amplifies most. AI is a brilliant specialist; it can go deep on almost anything. What it can't do is zoom out and understand the full picture. That's where experienced generalists thrive, and it's where ex-managers have a genuine edge.

The other thing management gave me that AI simply can't replicate is understanding the why behind what we're building. Years of working with stakeholders, translating business problems into technical solutions, understanding which trade-offs actually matter to the business versus which ones are purely technical vanity, budget constraints. An engineer who genuinely understands the business context and can pair that with AI-assisted delivery velocity is a powerful combination. Gergely Orosz wrote an excellent article on Product-minded Engineers, which I think is essentially what a manager with AI becomes.

I believe an experienced generalist, who understands both the technical and commercial landscape, augmented by AI, is going to become increasingly valuable as our industry evolves.

The emotional reality

I'd be lying if I said this transition was all positive momentum and career satisfaction. There's an emotional side to it that I think most people considering this move don't fully anticipate.

Imposter syndrome is very real, and it hits differently when AI is involved. There are days when I feel like Claude is doing the real work and I'm just along for the ride. Like I'm not really an engineer, I'm just a manager who's good at prompting. That voice is persistent and it's surprisingly hard to shake.

The reality, which I have to keep reminding myself of, is that knowing what to build, how to architect it, and being able to critically evaluate whether the output is good enough, is the hard part. That's the bit that takes over a decade to develop. The AI is a tool, an incredibly powerful one, but the judgement and direction comes from experience. Asking for regular feedback from peers and managers has been really important for me in reinforcing that I do know what I’m doing!

Practical playbook

For anyone seriously considering making this transition, here's what I've learned about the practicalities.

It is almost certainly easier if you move to a different company. This is probably the single most important piece of advice. At your current business you'll always be the manager. People will still come to you with management problems. Former direct reports becoming peers creates awkward dynamics that are difficult to navigate for everyone involved. A clean break makes it much easier to establish yourself in a new capacity.

Target the right kind of role. Look for positions at smaller businesses or startups where your leadership experience is part of the value you bring, not something you have to explain away. Think first tech hire, Founding Engineer, Principal Engineer building out a new department. These roles actively benefit from someone who can think strategically and lead when needed, as well as write code. They also tend to mean you don't have to take a significant pay cut, because you're offering something that a pure software engineer candidate can't.

Leverage your network. This transition is significantly harder to make through cold applications. A hiring manager scanning CVs on LinkedIn is going to see a recent history of management titles and move on. You need people who can tell your story, recruiters who understand what you're trying to do, former colleagues who can vouch for your technical ability, connections who can get you a conversation rather than just an application in a pile. I was very fortunate to work with a recruiter who understood my situation and could advocate for me directly with hiring managers. That made all the difference.

Be strategic about interview processes. This is crucial. Seek out companies that interview based on real-world thinking, technical conversation, and problem-solving approach rather than LeetCode-style coding tests. To be completely honest, I would have struggled with a traditional tech test. The interview for my current role was a conversation about my experience, my technical approach, and how I think about systems. That's where ex-managers shine, and it's a much better indicator of how someone will actually perform in the role anyway (don’t get me started, that’s another article coming soon!)

The pendulum keeps swinging

A few months into this transition, I can say with confidence that the time I spent in leadership has made me a significantly better engineer. Not in spite of the years away from the tools, but because of everything those years taught me.

And the engineering work is already making me a better leader. Being close to the tools, understanding what AI can and can't do in practice, seeing first-hand how modern engineering teams can operate. All of that context is invaluable for anyone making technical strategy decisions.

I think AI is going to make the Manager/Engineer Pendulum much more accessible. The ramp-up time that used to make the swing back to engineering feel impossible has been dramatically reduced. We're going to see more senior technical leaders moving fluidly between leadership and hands-on work, rather than treating management as a one-way promotion. I know I’ll certainly be looking to swing back and forth in the future.

For anyone in a leadership role feeling that pull back to engineering, or any engineer wondering whether a move into management means leaving the tools behind forever, it doesn't have to be a permanent choice. The pendulum can swing both ways, and the people who embrace both sides of it are going to be the ones who thrive in this new era.