No lag, no drama. Backend, infra, and the bits in between.
The longer version, in case you’re curious.
I joined CultureX as the founding engineer back when I was an intern still figuring out what Node.js was. We built the product from scratch. Today it’s used by 150+ enterprise teams (Mahindra, Nestlé, Decathlon among them) to run influencer marketing at a scale that didn’t exist when we started.
I’m not the engineer who works 24/7. I’m the one who likes understanding tech deeply, exploring products, listening to user feedback, and being able to talk about all of it without burying anyone in jargon.
A developer who loves product. That’s the shortest honest version. I care about the user feedback loop as much as the system architecture, and I think the best engineers are the ones who can hold both in their head at once.
Lately I’m thinking a lot about how AI is reshaping how we build products. Not the demo-day version: the boring useful version that quietly makes someone’s day a little easier.
What I actually reach for.
Most days I’m in the backend: Node, TypeScript, Mongo, Postgres, queues that occasionally misbehave. I like understanding systems from the deep core: how they actually work, not just how to use them.
I’ve grown into the infra layer too: AWS Lambda, EC2, SQS, the moment a 90-second API becomes sub-second. The unglamorous bits, but the ones that make everything else possible.
I keep one foot in the frontend so I never forget what the user actually sees. React, Next.js, the difference between functional and feels-good. That’s the bar.
A few problems I’m proud of having sat with.
The old MongoDB schema was working but groaning. I rewrote it for the way we actually queried, and migrated 8M+ docs to the new shape during a single 8-hour cutover. No downtime, queries 70% faster the morning after.
8M+ documents migrated, zero downtime.
The SuperAdmin analytics dashboard took 90+ seconds to load. The fun of this one was the detective work: query profiling, finding the indexes the database actually wanted, deleting code that turned out to be the problem.
99% faster. Same data.
I lead a team of ten engineers now. The work I’ve enjoyed most isn’t the architecture diagrams. It’s the back-and-forth in code reviews, learning how juniors and seniors think about a problem and adjusting how I think about it.
Founding engineer → engineering lead. ~300 reviews, mentor to many.
And where I’ve done it.
The complete version, with dates and dependencies, is in the résumé (pdf).
A couple of things I’ve built for myself.
Built to scratch a personal itch. The kind of project where you find out how a thing actually works by making one. FastAPI, Anthropic’s MCP, deployed somewhere I can forget about.
Notes from along the way.
Mostly on the boring useful corners of building products. Full archive at dipeshjaiswal.com/writing.
I want to build software that disappears into people’s lives. Quietly useful, almost boring in how reliable it is. AI as an ingredient, not the headline.
If any of this resonates, say hi.
Email is the easiest. I read everything.