Junior Engineer Built a Tool That Got Him Promoted Over a Coworker Who Was His Friend — It Sparked a Falling Out Between Them

Photo credit: AI-generated image created using ChatGPT. Illustrative only.

Eight months into his first industry job, a 25-year-old junior engineer decided the work felt slow and small compared to the AI research he’d just finished in grad school. So he started building something on the side—quietly, on his own time—without telling his team.

When the tool started working, it didn’t just get attention. It changed his trajectory at the company, and it also changed one workplace friendship in a way he didn’t expect, according to the original post.

He wanted a bigger role, and kept hitting a wall

The engineer said he joined a mid-sized tech company with roughly 4,000 employees, most of them in software. Despite the scale, his day-to-day work felt “pretty boring” compared to what he’d been doing in academia.

He tried to jump ahead—applying for senior and staff roles—only to get rejected repeatedly. He said the only places that seemed interested were early-stage startups, and that mismatch left him frustrated and restless.

Instead of waiting for a new opening or a formal project assignment, he went looking for a way to prove he could deliver something bigger.

The side project was an AI “scoreboard” for code quality

On his own time, he began building an AI tool he kept secret from coworkers, not because it violated policy (he didn’t describe any rule-breaking), but because he worried people would tell him to stop or call it a distraction.

He trained the system using public GitHub repositories at first. The tool’s job was to identify engineering patterns and measure things like code clarity regressions, architecture choices, and whether contributions followed best practices. It would then generate a score.

It wasn’t only backward-looking. He said it could offer advice before someone pushed new contributions, positioning it as coaching rather than punishment. But the feature that made it immediately socially explosive was the leaderboard—tracking who consistently wrote “clean stable stuff” and who tended to introduce issues that needed later fixes.

Then he showed leadership, and it escalated fast

Once he felt the tool was solid, he brought it to the director of his organization. The director liked it and told him to test it on the company’s internal codebase, since most code was visible internally.

That test apparently went even better than expected. The engineer said he was promoted to tech lead for the tool, and teams began adopting it. Managers saw value in the explanations for both evaluation and helping people grow, and he heard from others across the org that it made workflows smoother.

At the time he wrote the post, the company was preparing to introduce it more broadly, potentially across other organizations. For someone eight months into a junior role, it was a rare kind of rocket ride: a self-started project turning into an internal platform with a formal leadership title attached.

A former teammate’s ranking turned it into something personal

The personal fallout centered on one coworker from his old team, a 27-year-old woman he described as someone he’d been friendly with—coffee chats, casual conversation, the kind of everyday workplace familiarity that feels stable until it isn’t.

After the tool rolled out, her behavior toward him shifted. He said she began glaring at him or ignoring him entirely. He connected the change to the leaderboard: she was ranked fourth from the bottom, and the dashboard could be sorted so that low performers were easy to spot.

He insisted he never singled her out—no comments, no jokes, no public callouts. He also said there had been no discipline tied to the tool: no one punished, fired, or even scolded because of a score.

To him, the system was meant to show patterns and encourage improvement, with a disclaimer on the dashboard saying it wasn’t designed for shaming or blame. He also pointed out a nuance he felt mattered: just because someone committed code doesn’t necessarily mean they chose the approach, since they might have been following a superior’s instructions.

But from her side, the results were now sitting in a place where others could see them. Even if no manager used it punitively, the ranking alone created a new label—one she may have felt stuck with in front of peers.

Why the tool’s “helpful” design can still sting

The engineer framed his intent as improvement and analysis, and the company response he described suggests leadership saw it the same way. Still, the mechanics of a leaderboard can turn “feedback” into status, instantly and publicly.

A score that follows an employee around—especially one tied to something as identity-heavy as code quality—can feel less like coaching and more like a scarlet letter. Even if the tool is accurate, being near the bottom creates social consequences that don’t require official punishment: coworkers’ perceptions shift, trust can erode, and collaboration can feel riskier.

And in this case, there’s an extra twist: the tool didn’t just measure others. It helped elevate its creator. The engineer wasn’t simply the messenger delivering numbers; he became the new tech lead for the system that ranked his peers, including someone who had previously been his friend.

That mismatch—her visibility as a low-ranked contributor paired with his visible reward—may have been enough to make the friendship feel impossible to maintain, even if he never intended it that way.

People focused on workplace dynamics, not just intention

In his telling, the core question wasn’t whether the tool worked. It did, and leadership endorsed it. The question was whether building it—and accepting the promotion that came with it—made him the bad actor when a coworker felt humiliated.

Reactions to stories like this often hinge on the gap between intent and impact. A person can build something to be “just data” and still create a new hierarchy overnight, especially when the data is comparative and sortable.

Another practical point people tend to home in on in workplace conflicts like this is what happens next: whether management will keep the tool strictly developmental, whether access to rankings will be restricted, and whether employees will have a formal way to contest or contextualize scores—particularly when code quality reflects deadlines, legacy systems, or direction from senior engineers.

For the engineer himself, the immediate problem is smaller but thornier: how to address a friendship that has turned cold without making it worse, and how to keep a tool billed as supportive from becoming an unofficial scoreboard that teams start whispering about.

He said he never built it to embarrass anyone, and he didn’t think about who would land at the top or bottom until after it was deployed. But once the rankings went live, the social reality changed. And even with a disclaimer on the dashboard, one coworker’s reaction made it clear the tool had started measuring more than code.