We’ve just launched an AI Engineer role, and Pete’s written a post explaining why we’ve created the role which goes nicely with our blog series “Building with AI” that the team released the other week.
With all the noise in the industry right now, you might think a role like ‘AI Engineering’ is for SEO or an attempt to ride the hype wave. That’s not why we did it, though: we created it because the work is materially different, and we want to represent that clearly so people can make informed decisions when applying.
While Pete’s post explains what AI engineering means for us as a company, I want to address what this role might mean for you as an engineer considering the switch. If you’re a software engineer with an interest in AI thinking about making this pivot, what does that mean for your day-to-day work? Your career trajectory? What new experiences will you gain, and what familiar aspects of your current role might you leave behind?
This post is my honest pros-and-cons list that I’d give to a friend considering a switch into AI Engineering. I’ll be candid about both the exciting opportunities and the potential challenges, drawing from my own experience making a similar career transition.
First though, I want to share a personal story about where I found this kind of advice particularly valuable in my own career.
SRE
A year into my role at GoCardless working on payment pipelines, I faced a similar career decision. Our SRE team was short-staffed, and Norberto—then SRE lead, now VP Eng at incident.io—approached me about making the switch.
What I appreciated most about Norberto’s approach was his honesty. He emphasized this would be a fundamentally different experience than my current role. It wasn’t a lateral move or small adjustment, it was a significant career decision that would give me different growth opportunities and require commitment to see the payoff. Not a decision to make lightly.
His “pros” were compelling: I’d gain exposure to a huge amount of software by learning to run it, which would develop my sense for building better systems (absolutely true in retrospect). I’d build incident response skills (you can see how that ended up shaping my career!) and encounter more challenging technical areas—like distributed systems—than I might otherwise.
But it was his candid “anti-sell” points that I found most valuable:
- “Your development as a software engineer will slow, as you’ll write much less code.”
- “The work is further from the product, which changes your relationship to the business and understanding of the company.”
- “Infrastructure work can be long, tedious, extremely frustrating: if you’re not up for feedback loops that would break a normal dev, it may not be for you.”
I ended up working as an SRE for half a decade, eventually becoming Principal SRE and leading the infrastructure function at GoCardless. Norberto’s assessment couldn’t have been more accurate. Having that honest preview helped me make the most of my time in the role, and that experience became a key part of why I’m good at my job today.
I want to offer you the same candid assessment about AI Engineering.
Why switch to an ‘AI Engineer’ role?
Norberto’s advice proved invaluable because it focused on how I would experience the role as an individual—the unvarnished reality of the hard parts, what growth I’d gain, and what I might sacrifice. It helped me make the decision intentionally, and when I eventually hit the harder parts, I was prepared and ready to get through them.
I want to offer the same service for anyone considering our AI Engineer role. The ideal candidate is likely already in a Product Engineering position or something similar. So I’m going to share some “ground truths” about the experience you’d have working as an AI Engineer, particularly how it contrasts with typical Product Engineering work.
If you’re weighing this career move, here’s what you should consider:
It’s the number one priority
Most companies are racing to capture their AI opportunity, which makes AI work the company ‘P1’.
Working on a high-stakes, high-reward company initiative is a great way to get exposure to leadership and learn how to move fast under pressure. You’ll be at the center of strategic conversations, with visibility to executives and decision-makers who care deeply about your team’s success.
When I was in SRE, infrastructure projects often felt distant from company priorities until something broke. In contrast, AI work today is the key strategic direction for most companies, it’s what everyone wants to know about, and what many hopes are riding on.
If that sounds exciting, AI Engineering could be an opportunity to really push yourself and develop crucial skills in delivering under pressure. Of course, the reverse is also true: if you prefer steadier, more predictable work with less scrutiny, AI might not be the right fit, at least not for now while the race is on.
Potential for impact is huge
Until now, the products being built with AI were simply not able to exist.
That means a vast amount of potential value has become accessible and every company is starting the race at the same time. It’s an unusually level playing field that gives even small companies the opportunity to win.
In our case, if we manage to build an AI incident investigator that automates the debugging of an incident, that will fundamentally change how people respond to incidents. It will be a ‘before/after incident.io’ moment, representing a substantial shift in how the industry handles critical failures.
Opportunities to build category-defining products don’t come along often. If you want to be part of building something truly game-changing and associate yourself with a product that people will remember as being the first of its kind, then an AI Engineering role is likely your best shot right now.
Help define best practices and tooling
You’ll be working on the forefront of a new tech wave, which means you have the opportunity to be the ‘first’ to learn how to do things, and to teach others.
We’ve seen this already at incident.io, where we’ve had to create tooling from scratch to help us build our investigations system. Everyone in the industry is trying to solve these problems, but we’re among the first companies to build appropriate internal tools to address them. We’re not just implementing existing solutions, we’re inventing them.
What you build, you can share externally. You can see our “Building with AI” series where we share in detail how we work with AI systems. This kind of content helps build your personal brand as someone helping the industry figure out how to work with this new technology.
It’s not just blogs either, with our team speaking at conferences and regularly talking with other world-class teams about what’s working, what doesn’t.
Becoming an expert in a domain that is growing this fast, in a role where your company supports you speaking externally, is an exceptional way to grow yourself and your career.
Progress can be slow and inconsistent
Building AI systems comes with a unique set of challenges that traditional software engineering doesn’t prepare you for. Progress rarely follows a predictable, linear path—and this is both the hardest part of the job and where the greatest growth happens.
In traditional product engineering, you generally know when a feature is “done.” The requirements are clear, the acceptance criteria are defined, and you can check boxes as you go. In AI engineering, the landscape shifts constantly:
- Some weeks you’ll make remarkable progress, others you’ll feel like you’re moving backward
- Knowing when you’re “done” is hard—there’s always another edge case to handle or percentage point of accuracy to chase
- You’re working with probabilities instead of certainties, which fundamentally changes how you measure success
- There are few established patterns to follow; you’re creating them as you go
This uncertainty can be mentally taxing. The dopamine hit is way less reliable than smashing out product features—don’t do this if you live for consistent todo list checkboxes!
But this challenging environment is precisely where you’ll develop skills that are incredibly valuable and transferable:
- You’ll learn to define meaningful metrics that actually reflect real-world performance
- You’ll build tooling to measure quality and progress, turning the subjective into the objective
- You’ll develop grit and resilience that will serve you in any future role
- You’ll become comfortable making decisions with incomplete information
It’s not possible to “vibes” your way through AI engineering. You can’t just say a feature is good because it feels right, you have to build the datasets and tools to prove it with numbers. This rigor forces you to become a more disciplined engineer and project manager.
Taking on this challenge can help develop an ability to drive progress through ambiguity, which is one of the most differentiated skills you can have, either as an engineer or leader. This makes people exceptionally valuable not just in AI, but in any complex, uncertain engineering environment.
Product surface will be smaller
One significant shift when moving from product engineering to AI engineering is the visible impact of your work. Let’s be direct: you won’t build as much user-facing product surface as you would in a traditional product role.
In product engineering, you regularly ship features users directly interact with. You work closely with designers to build user facing interfaces, and you get immediate feedback when customers use what you’ve built. It’s satisfying to point at a screen and say, “I built that.”
AI engineering work is different:
- The majority of your work will be improving the system that powers the product surface, potentially leading to no visible changes
- You’ll likely work less with designers, as there is less user-facing parts of the product to be designed
- The “wow” moments come from how well the system is built under-the-hood, rather than visual polish
- Success is measured more in metrics and capabilities than a human-made judgement on quality
So less conventional software engineering work, but it’s not all bad: AI systems require extensive internal tooling, and as a person building the system and the intended of that tooling, you’ll have opportunities to extend those tools yourself. Lisa shows a lot of what we’ve built in “Why we built our own AI tooling” if you’re interested.
There’s a creative freedom in building internal tooling that is rare in customer-facing product. That’s because internal tools are most important to be functional, and the bar for polish is lower when shipping to an internal team than external, allowing you to quickly prototype experiences and even ship autonomously without any extensive design review.
It can be fun to build tools you use yourself, and to do so quickly, solving a problem that you are your team are personally impacted by.
It’s different enough
Just like SRE, AI Engineering is a different enough experience than Product Engineering that it’s worth giving it a name to make those differences clear.
The ideal candidates are generalist software engineers, but they need to be excited about the challenges this role presents. You want people who find AI technology genuinely fascinating, who enjoy keeping up with the latest advancements, and who won’t be frustrated by extended periods of gradually tweaking systems to improve on benchmarks.
This isn’t a role for everyone, and that’s okay. Some engineers thrive in the predictable rhythm of product development, where work is more predictable and feedback immediate. Others will find the frontier nature of AI work energising and might be looking for a career defining opportunity, leveraging the AI wave to see what they can achieve.
If you’re an engineer who reads this and thinks “wow, that sounds exciting!” then I urge you to consider making the switch. There’s so much opportunity in AI right now that it could be a huge career-accelerant, one that you’d look back on in future as a defining moment in your professional journey.
This is a chance to build something totally new and be recognised for getting there first, and a chance to help define AI Engineering as a discipline. The skills you develop working with these systems today will be increasingly valuable as AI continues to transform how we build software.
If that’s you, I’d love to chat. Either reach out to me on LinkedIn, email me or apply directly at incident.io/careers. Even if you’re just curious and not ready to apply, I’m happy to discuss what the transition might look like for someone with your background.
If you liked this post and want to see more, follow me at @lawrjones.