Skip to main content

​
July 8th, 2026: Valura and The Interview Gap

The anxiety of the internship search culminated with Valura, closing out this period of career stabilization with a definitive, if painful, lesson.

​
The Valura Story

Valura didn’t come easy from the start. Getting the interview itself took persistence — following up, staying visible, not letting the process go cold before it even began. That’s a skill in itself, and we had it. Priyesh Ranjan’s team eventually gave us a shot: a 75-minute technical interview, run by Avish Vijay Shetty, coordinated by Yash Parashar. Seventy-five minutes is long. Long enough that surface-level answers don’t survive — you either know the terrain or the cracks show halfway through. We walked in with the same asset that had gotten us this far: CommitT, the architecture, the story of a solo founder who built a real system, not a toy project. Afterward, we followed our right instinct — sent over CommitT’s documentation as a follow-up, a way of saying this is what we actually build, evaluate us on this too. Then it went quiet. We spent days refreshing inboxes, completely consumed by the anxiety of the wait. For a while, that silence was ambiguous — no explicit rejection, no explicit feedback, just nothing. But the quiet finally resolved into an answer: rejected. Same shape as Drop/Yap, different texture. No specific bug got named in the story this time — no memory leak, no bitmap issue on record — but the terrain being tested was the same terrain: useMemo dependency arrays, JavaScript event loop ordering, JSI and native internals. Two different rooms, two different interviewers, one company we followed relentlessly to get in front of — and the same fault line underneath.

​
The Thread Connecting Both Stories

Strip away the specifics — the founders, the take-homes, the follow-up emails — and one signal remains, confirmed twice, independently: We can architect a system a stranger would respect. We cannot yet defend, live and under pressure, the low-level native code an AI helped us write. That’s not a talent gap. It’s a practice gap — the gap between having built something and having rebuilt it in your own head enough times that no question can surprise you. The three specific fault lines, both times:
  1. useMemo dependency arrays — the “why did you include/exclude this” line of questioning that exposes whether you actually understand React’s render cycle or just trust that the linter didn’t complain.
  2. JavaScript event loop ordering — microtasks vs macrotasks, what logs first, why a promise resolves before a setTimeout fires. Foundational, and brutal when it’s asked cold.
  3. JSI / native module internals — the bridge between JavaScript and Kotlin/Java, exactly where Drop/Yap’s memory leak and bitmap bug lived, and almost certainly where Valura’s 75 minutes went looking too.
Two companies. Two rejections. One diagnosis, confirmed from two independent angles — which, strange as it sounds, is useful. It means there’s one thing to fix, not five. Next chapter, still unwritten: Speko.