The long way around.
Mechanical Engineering degree. Self-taught developer. A startup that took three years and everything I had. A job that didn't fit. Then the one that did.
Gaming — before I knew what I was learning
I grew up on mobile and console adventure games. The kind where you notice the level design before the story. I'd finish a game and immediately start thinking about what I'd change — the pacing, the layout, the rules. I didn't know that was design thinking. I just thought it was being annoying.
First PC, first builds
First year of engineering. First real machine: HP Pavilion Gaming Laptop. I installed everything I'd heard people use — Photoshop, Premiere Pro, After Effects — not for any project, just to see what they did. Picked up Monkeytype, got past 100 WPM. Then installed Unity, taught myself C#, built an arcade-style game with hand-drawn Photoshop assets. No course, no bootcamp. The game shipped. I learned that I could pick up whatever I needed if I were willing to be confused for a while.
A startup, accidentally
A friend saw my Photoshop work and saw a market. We started selling custom bike wraps — I handled design, he handled business. When that got repetitive, I wrote JSX automation scripts to generate mockups. Then the business needed a website. Hired a developer — didn't work out. So I spent a few weeks learning HTML, CSS, JavaScript, React, then Next.js. Built the site myself. Debugging production issues alone at 2 AM with nobody to ask was an education in architecture, error handling, and reading documentation properly.
Building a real platform
Window pillar wraps broke through. The team grew to ten. I became the sole engineer behind the entire platform. The vision shifted constantly — 3D garage configurator, marketplace, custom offer engine, B2B — so the system had to be flexible enough to support whatever came next. The admin panel ended up handling inventory management, multi-domain management, offer engines, customer journey analysis, A/B testing, review moderation, personalised fulfilment flows, and multi-store shipping. It had most of what Shopify offers a single store — but built around our specific workflows and designed to feel simple.
Infra at scale, on a budget
60,000+ monthly users. AWS bill under $2 (S3 + CloudFront). MongoDB under $9/month. Vercel at $20/month. Getting there wasn't straightforward — I spent weeks finding the right config for MongoDB connection pooling on Vercel's serverless architecture. Payment failures from Razorpay's upstream bank downtimes led me to build multi-orchestration with PayU as a fallback. Then Vercel raised their pricing — our bill jumped to $116, optimized down to $75, but the team decided to migrate to Shopify. Grow faster, spend less, fewer payment incidents. It was the right business call.
Voltas — the wrong fit
After graduating, I joined Voltas as a billing engineer. It was the kind of role where you hold your head with a headache every day at a desk doing work that has nothing to do with what you're good at. I'd code during lunch breaks. After hours, I'd work on MaddyCustom. I was actively looking for a software engineering role, but a Mechanical Engineering degree and a weak public portfolio made it harder than it should have been.
Blitzit — the right one
Blitzit gave me a take-home assignment. I solved it. Did the interview. Got the offer. Now I work as a full-stack software developer — the kind of work I'd been doing for years, except now it's the job title too. I integrated Asana's two-way sync from scratch, built an MCP server, implemented a notification system with BullMQ and Redis queues, debugged and improved the Notion and Google Calendar integrations, and continue to ship complex features every week.
The exit
I wasn't interested in managing a Shopify store. What I cared about was building — the systems, the logic, the architecture decisions. So I stepped away from MaddyCustom. It was three years of shipping under real constraints — real users, real money, real outages. Nights of two hours of sleep. Trade-offs between grades and uptime. I leaned in harder than was probably healthy. But it made me the engineer I am now, and I don't regret the exchange.
What I'm doing right now
At Blitzit, I'm building a new, more robust backend from scratch — unit-tested, modular, plugin-based integrations, auto-switching model routing with prompt-level control, personalised Pinecone vector memory, and tighter security: rate limiting, request validation, and defences against the usual attack vectors. Outside of work, I'm learning data science and machine learning, and starting to explore deep learning. On the side, Spyll — the anonymous social platform for college students — hit 1,000+ Android downloads in its first month after launch, with zero advertising. Now I'm expanding it.