About

I started a B.Tech in computer science at NIT Kurukshetra in November 2022. In my second year I started SynthAILabs (first called OpenEdu): free, AI-made learning material built around projects, for engineering students who had the ability but not the ramp.

It grew into a community of 11,000+ students across India. Alongside it I co-founded OpenCode Developers, a club that took 500+ students through competitive programming in one summer break instead of the usual six months. Some of them later placed at Morgan Stanley, JP Morgan and Google.

I wanted SynthAILabs to stay free and non-profit, and I couldn't fund that, so I wound it down. It was my first lesson that a thing people want and a thing that pays for itself are different problems.

In December 2024 I dropped out to join AuraX as AI researcher and CTO, with Rahul Khatri. We built image models for fashion and product brands: virtual try-on and base image generation, incubated at IIIT Hyderabad's CIE.

The core was an aesthetic fine-tuning loop: generate an image, score it for aesthetics and garment accuracy, feed the scores back, repeat. In blind tests people picked our images as real over Google's and OpenAI's image models. On Nautinati, Aditya Birla Fashion's kidswear brand, our images got a 3% click-through rate, against 1.5% for studio shoots.

We raised ₹22 lakh from IIIT Hyderabad's CIE, IIM Bangalore's NSRCEL and MeitY's TIDE 2.0, and closed deals with Mensa Brands, Aditya Birla Fashion, TMRW, SKORE and TTK Healthcare.

Then the next round didn't come in time, and the big labs moved in. We turned down ₹28.5 lakh more: it would have kept us alive, but not ahead. The whole story is here.

After AuraX closed I took time off. I focused on health, which I had ignored for two years of building, and studied finance to understand the business mistakes I'd made.

That turned into Qubit Quant Global: a lab that tested 480 rule-based strategy ideas on real tick data and kept the few that held up on out-of-sample years. They run as MetaTrader 5 Expert Advisors. Out of it came Qubit Trade Manager, an MT5 trading panel with 600+ downloads on MQL5 and no marketing. I still run it on the side, at qubitquantglobal.com.

For five months I was a software and AI developer at Elip AI (then Tejas AI, YC W25): WhatsApp agents with memory across sessions, a scraping fleet indexing about 60 ATS platforms, and OpenSearch hybrid retrieval doing the matching.

None of the hard parts were model problems. They were memory, deterministic routing, and knowing when a requirement is a filter rather than a similarity score.

I left in August 2026 to build my own studio full time.

AuraX had the better product and still ran out of time. The part we never got good at was getting it to people. Marl Studio is built around that: launch a product, work out its go-to-market, grow it, then do it again.

The first products are Bay and Bay Effects, free AI video and motion graphics editors for Mac, Windows and Linux, rebuilt on ashlar, my Rust app framework. They ship a release every two to three days. Next is hull, a game engine, with Shroud as its first game.

Outside work: 1,800+ books, two national-level school basketball golds, and night photos.