Software engineer building products people can rely on.
I’m Nikhil Verma, and I’m based in Mumbai. I like taking an idea the whole way: understanding the problem, building the system behind it, and caring about how it feels to the person on the other side.
Right now that’s Trailgrad, a place for engineers to prepare for interviews. Before that, I spent two years at Flashaid, working on payments and insurance.
Trailgrad
Preparing for technical interviews is mostly done alone, and it rarely feels like the real thing.
Trailgrad is my attempt to change that. It’s a place where engineers can practise the way interviews actually happen: thinking out loud, being asked the follow-up question, and getting honest feedback afterwards.
It starts from the person, not a question bank. What you’ve built, where you want to go, and what you find hard shape what you practise next. And when you’re stuck, there’s always a next step, whether that’s a nudge, an AI interviewer to rehearse with, or another engineer who has solved the same problem.
Good preparation should feel less like a test, and more like having someone in your corner.
I’ve built Trailgrad on my own, from the product decisions to the infrastructure. A lot of the interesting work is the part nobody sees: keeping a live voice conversation responsive, making sure feedback is fair and consistent, and keeping things reliable when a lot is happening at once.
Next.js · TypeScript · PostgreSQL · Redis · Gemini · Groq · LiveKit · Yjs
Flashaid
Flashaid was where I learned what it means to ship software people depend on. I worked on the products behind its payments and insurance, the systems that quietly need to work every single day.
Most of my time went into SmartCollect, Flashaid’s payment collections platform. I built much of it across the backend, the frontend, and the payment gateway integration, and it now handles more than two thousand payments a month.
Payments taught me to care about the things users never see: what happens when a request arrives twice, when a provider is slow, or when something fails halfway through. I still build that way.
I also worked closely with the business side, turning what they needed into features for the insurance products, and helped bring on partners that took those products to students across dozens of institutes.
How I work
I write mostly TypeScript, with React and Next.js on the front and Node.js, PostgreSQL, and Redis behind it. Lately much of my work has been with language models and voice AI: wiring them into products, checking their output, and making them dependable enough for real users.
TypeScript · JavaScript · Java · SQL · React · Next.js · Node.js · Tailwind CSS · PostgreSQL · Redis · Prisma · WebSockets · LLMs and RAG · Voice AI · AWS · Vercel · Docker
Background
Say hello
If you’re building something interesting, hiring, or just want to talk about making good products, I’d be glad to hear from you.