This as-told-to essay is based on a conversation with Maahir Sharma, a 24-year-old software engineer in Dublin, Ireland. His employment and identity have been verified by Business Insider. The following has been edited for length and clarity.
Loading audio narration…
I work in a division that builds internal AI tools.
A year and a half ago, I used AI to help solve small bugs, like a UI fix or adjusting the margin of a website. Now, I’m using it to ship full-fledged production features and projects. I treat AI as a junior-level engineer with whom I have to collaborate.
I started coding when I was in fifth grade. When AI first started taking over my coding tasks in the last few months, it felt like a loss because I spent years trying to master this art — but I always had a feeling that the art would get redundant.
Initially, it felt like AI was taking over my job. But now, as I transition from a junior to a senior engineering role, I understand that the job was never about writing code.
I’ve transitioned to a product engineer
As a junior engineer, my original job was to get requirements, write code, build the feature, and ship it to production. I used to assume what customers wanted and would build something and ship it to production, but there were no actual users for it.
I’m no longer getting a clearly scoped set of requirements with the goal of shipping code. I feel like a mini business owner and product engineer. This transition has helped me understand how startups and entrepreneurship work.
Now it’s more about understanding the product, what customers want, the end-to-end business ecosystem, and then building a product.
Rather than just focus on the solution, I try to break down a problem into smaller sub-problems, and then I use my AI-assisted coding tool to generate a three-step plan.
First, I map out a plan with AI to figure out what we want to build, prompting it to ask clarifying questions, and map out the dependencies and features which we would want to see in the final product.
I keep on iterating on the prompt, usually until the eighth or ninth version, where I’ve arrived at a plan that considers all the technical decisions, optimizations, and tradeoffs, and want AI to generate the final code.
Then sub-agents start working on various features, and I closely monitor all the code being generated and review it line by line, correcting it if it starts to deviate. I think that’s the most important part.
Finally, I ask the AI to run extensive tests. I also ask AI to scan the codebase for any obvious security vulnerabilities and fix those. This is a step that a lot of people miss. You have to treat AI as a collaborator, not a code-generation machine.
Sometimes AI over-engineers the problem by throwing infrastructure components at it, so I try to determine whether the components it adds are actually required or if there’s a simpler solution.
I’ve had to shift my skillset
When I first entered the industry, I wasn’t really focused on soft skills. I used to speak a lot, but with less clarity. I think the transition has helped me communicate clearly with stakeholders and conduct my own market research on a product before building a particular feature.
I started using the STAR format when communicating with stakeholders, which stands for: Situation, Task, Action, and Result. If I’m trying to understand a business problem, I try to understand the situation, the task at hand, what kind of actions we can take, and the end results that we want in terms of business metrics or acceptance criteria.
I didn’t have this kind of product mindset six months ago. I used to focus on a lot of technical buzzwords. I would explain technical solutions to people who didn’t understand them. Being able to effectively communicate technical concepts is something I’ve learned over the past six months.
I feel optimistic about my future
I’m learning AI every day, and I spend roughly 20 hours a week upskilling through online courses or by experimenting with tools.
I’m enrolled in a lot of courses outside work in order to keep my skills up-to-date because this industry is transitioning to an AI engineering industry. People want intelligent workflows, and I think that’s something you need to keep up with as new technologies are released.
I’m optimistic about my job prospects in the next five to 10 years because I spend time on the right things.
