Interviewing as a software engineer has never had a real playbook. The process is different at every company, it is genuinely hard on both sides of the table, and the one thing the industry has mostly standardized on, the LeetCode-style algorithm round, won that spot by being objective and easy to compare across candidates, even though it is an artificial test with very little connection to the actual work.
I have job-searched the normal way before, and I knew what that playbook returned. Cold applications disappeared into silence, no reply, no rejection, nothing. Referrals did better, they got me into rooms, but the rooms ended in interviews that ended in nothing. And the track record was only half of it. Cold-applying into silence is quietly demoralizing, every unanswered application chips away at you a little, and grinding LeetCode is a kind of studying I have never enjoyed. What I remember most is that the whole process left me feeling worse every week I kept it up.
So this time I tried close to the opposite. I did not grind LeetCode, and I did not send a single cold application. Since the start of the year I have had real conversations with six companies, every one of them inbound, and two of them became offers.
The bet
The bet was simple to state and slow to execute. Instead of preparing to look good in interviews, I would spend months being visibly good at the actual work, in public, where anyone could see it.
What made the bet easy to commit to was that it paid off no matter what happened. Writing essays, contributing to open source, building things in public, all of it makes me a better engineer and leaves something real behind, whether or not it ever turned into a single interview. I was choosing a path that was good for my career in the worst case, and that I genuinely enjoyed doing. If the search produced nothing, I would still have spent that time getting better at the thing I do. The offers were the upside. They were not the point.
Concretely that meant a few things. I wrote roughly an essay a week, about two dozen of them so far this year, on engineering and AI and the things I was actually doing. I built in the open, which this year came to 146 merged pull requests across 45 repositories, including projects like Ink and Homebrew and Directus, real codebases with real maintainers and real review standards. I built a personal site that updates itself, so every new essay and every new contribution shows up without me touching it, and the site is never stale. I posted on LinkedIn here and there, lightly.
And there was a piece of standard advice I skipped completely. I never asked my network for referrals, even though working your contacts is the first thing everyone tells you to do.
Spending the leverage
There is a second half to this that matters as much as the visibility, which is that being visible gave me leverage, and I spent the leverage on being selective.
I was deliberate about the kind of interview process I would say yes to. I steered toward work trials, architecture discussions, and take-home challenges, the formats where you do something close to the real job, and I declined loops built around LeetCode-style algorithm rounds. When the inbound interest is steady, you do not have to accept every format. You can pick.
I was also deliberate about which companies. I targeted companies that allow and actively champion AI-assisted development, because that is the work I want to do, and I am not interested in going back to coding the manual way. That filter cut the field down fast, and it cut it in a direction I actually cared about.
What came back
There were six of these conversations all told, and every one of them started as an inbound contact, a recruiter or a referral that found me. Zero of them started with an application I sent.
It was not fast, and that turned out not to matter, because the inbound was never the number I was watching. For the first few months almost nothing came in. I was doing this work because it made me better and I enjoyed it, and the inbound interest growing was a side effect I had not planned on. If it had never come at all, the time still would not have been wasted. I would have gone into a proactive search, cold applications and all, with a real body of work to point to instead of starting from nothing.
Then the inbound did start, slowly, and then steadily. Of the six conversations, one company passed on me after a first round, which is a normal and healthy thing to have in a funnel and I am mentioning it on purpose. Two of them ended in offers. One came from a recruiter who found me, and the other, the one I accepted, came from a former colleague. We had worked well together, we had kept in touch since, and when a role came up that he thought suited me, he was the one who reached out. That relationship was the whole reason the conversation happened.
Walking through the door
The visibility opened the doors. Walking through them still took work, and I had a system for that part.
I keep one resume that is genuinely complete. It has everything, every role, every project, the leadership work and the hands-on engineering and the AI work, and it is far too long to send to anyone, five or six pages. It is not a document I submit. It is a source of truth. When a specific role comes up, I use AI to research the company and read the job description closely, then build a tailored resume from it, one or two pages that cherry-picks only the experience that matters for that role.
I also built a small set of strong stories over the course of the process, the kind that can be aimed at different roles depending on what each one needs, and I practiced them until they were smooth. Before each conversation I researched the people I would be talking to and the company itself, so the questions I asked were good ones.
None of that is novel on its own. What made it work was that the door was already open before I started preparing. Preparation is a much better use of energy when it is not also carrying the job of getting noticed.
The part I did not plan
This is the part I would keep if I had to throw the rest away.
Spending months writing, contributing to open source, and building things in public is not a job-search tactic that sits on top of your real job and competes with it for time. It is the same activity as getting better at the work itself. Reading unfamiliar codebases well enough to contribute to them, explaining your thinking clearly enough to publish it, shipping work that strangers will judge, that is just practice, and the practice compounds.
I felt it at my current job. The contributing and the writing made me a better engineer, which I expected. What I did not expect is that building my own projects, where I had to decide what was worth making and why, made me sharply more product-minded, and that fed straight into the product side of my role. In the same months I was doing all of this, I was exceeding expectations, and I was promoted. I did not split my attention between my job and my job search and do both at half strength. The thing I was doing for one turned out to be the thing that improved the other.
What I actually think this means
I want to be careful about how I land this, because I am not trying to tell anyone how to run their search. This is one person, one search, one result. The interview format I avoided works fine for plenty of people, and my slow inbound ramp could just as easily have stayed slow.
But the reframe is the part I believe in. A job search is usually a frantic, separate project you spin up when you need it and shut down when it is over. What I tried instead was being continuously legible, doing visible work all the time, so that there was nothing to spin up. The search stopped being an event. It became a side effect of the work I was already doing. And the next time I want to move, the work will already be done, because it is the same work.
The habit stuck
A month into the new job, and none of it has stopped. I am still writing at close to a weekly cadence, still contributing to open source, still building more side projects than I can finish. None of it is a search anymore. It just became how I work.
Part of that is how much fun it is to build with AI tools right now. The loop is fast, the ceiling is high, and it is a good time to be making things. But part of it is that the habit took. It stopped being something I decided to do each week and turned into part of my flow, closer to a hobby I can't put down than a plan I am executing.
