5 mistakes that cost me GSoC and LFX

Aswani Sahoo — Sun, 14 Jun 2026 11:03:06 +0000

It took me five rejections to understand this: I had the merged pull requests. I shipped real code into real repos for months. And I still went 0 for 5 across Google Summer of Code and LFX Mentorship.

So this isn't a "you just need to contribute more" post. Everyone already knows that, and in my case it was wrong. My code was fine. The five decisions I made before writing that code are what sank me. Those are the ones nobody warns you about.

If you're applying to GSoC or LFX, these are the mistakes I'd undo first.

Mistake 1: I poured months into an org that wasn't even playing

I picked my primary org early and went deep. Four months, starting from zero, ending with merged PRs in their codebase. Real relationships, real understanding of the code.

Then the accepted-orgs list came out, and they weren't on it that cycle.

Four months of work, pointed at an arena that wasn't open. It wasn't all lost: those months taught me how a real codebase and a real project actually run, and that grounding is exactly what let me spot a second org doing similar work and move there fast. So the skills transferred. The application credit didn't. Not one of those merged PRs counted toward the program itself. Check that your target org is accepted for the current cycle before you bet months on it.

Mistake 2: I bet on new orgs without counting the slots

After that fell through, I scrambled to two orgs that fit my domain. Both were running the program for the first time. Between them they had three and four project ideas listed.

I didn't think about what that meant. New orgs get fewer slots. Fewer ideas listed usually means fewer funded positions. I was competing for maybe one or two seats per org without realizing the board was that small.

Now I read the board before I sit down: how many ideas, how many slots the org got last year, how many applicants are already active. Pick arenas where the math isn't stacked against you before you write a word.

Mistake 3: I applied to a side idea, not the flagship

In one of those orgs I submitted a proposal for an idea I liked. I got a positive reply from a mentor, merged a couple of PRs, felt good about it.

Then the org got cut to a single slot. When that happens, the org funds its most important project first. Mine was a secondary idea, so it was the first to get dropped. Good proposal, wrong target.

Lesson: when slots are scarce, a strong proposal on a minor idea loses to an average one on the flagship. Aim at what the org cares most about, not what's most comfortable for you.

Mistake 4: I played the same game in every arena

This is the one I'd tattoo somewhere. Every org rewards something different, and they tell you what.

One org said plainly that they would weigh a thoughtful proposal over a pile of noisy PRs. So what did I do? I tried to out-ship the PR count anyway, then over-invested in analysis. The other program rewarded merged-PR velocity, and I showed up with a proof-of-concept and a polished cover letter instead.

I read the same playbook into every situation. Read the org's actual stated signal, then optimize for that one. Proposal-first orgs and code-first orgs are different sports.

Mistake 5: I didn't protect the window that mattered

For one application, the decisive two weeks landed on top of my mid-term exams. I lost about seven days right when momentum mattered most. In that gap, other applicants shipped sharp, relevant work and pulled ahead.

And when I did have time, I spent it refining strategy and analysis instead of shipping the one thing that would have moved me up. I was busy, not effective.

You can't always move the calendar. But you can know which two weeks decide the outcome and guard them, and when time is short, ship the smallest useful thing instead of polishing a plan.

The through-line

None of these is "write more code." Every one is a decision made before the code: which arena, how big the board, which idea, which signal to chase, which week to protect.

I was optimizing effort when the difference was sitting in the choices that come before effort. That's a fixable problem, which is the only good news in a 0-for-5 record.

Next post: the org I'm targeting next, and exactly what I'm doing differently this time. I'd rather show you the plan than tell you it worked.