From Nilphamari to the world.
The professional version is on the About page. This is the longer one — the failures, the family, and the slow build of building something that lasts. Read what you want.
Chapter 01
2008Where it started
I was a teenager in Nilphamari with a hand-me-down PC and an internet line that worked when it felt like it. I taught myself web design on dial-up. The first project anyone paid for was a Bangla wedding-invitation site for a neighbor — five hundred taka and a forever hook.
Chapter 02
2011NextBangla becomes real
At 20, I formalized the business as NextBangla Ltd. The first contracts were small — a logo here, a brochure there, a regional NGO's site. The bigger thing I learned that year was operations: invoicing, scope, contracts, client expectations. The work started looking like a business.
Chapter 03
2014The first failures
We tried to build a B2B marketplace. We had the best UX in our category and zero distribution. We tried an IT-staffing arm and shut it down rather than dilute the brand. Three other things didn't work either. Lessons cost real money. They're also why everything that came after worked.
Chapter 04
2018London calls
First UK clients. The work shifted — bigger budgets, longer engagements, calendar invites in two time zones. I learned how to defend a creative brief in writing. NextBangla started being mentioned in rooms I wasn't in.
Chapter 05
2020Joya joins as Managing Director
The partnership had been informal since the start. In 2020 it became formal: Joya took the MD role at NextBangla. Operations stopped being a thing I held together and started being a thing she ran. Everything since this date moves faster because of this date.
Chapter 06
2022Luxembourg
Worked with boskovGROUP Sàrl as Head of Design & Development. Three years of running production from a small office in Luxembourg taught me what European-grade ops actually looks like — and how much of that we could bring back to Dhaka without losing the cost arbitrage.
Chapter 07
202410 AI startups
The pivot. I'd been integrating AI into client work for a year. In 2024 I committed: 10 AI products, built in parallel, with the same team that ships client work. Voxly first, NOBBYO second, eight others on the bench. The experiment is whether the same principles that scaled NextBangla scale across products.
Chapter 08
2026Now — building #16
Six AI products live, four building. Client work is now exclusively founder-to-founder. The site you're reading is part of the same year-long push. If you're a founder ready to ship something AI-first, this is the era to talk.
The reason any of it works.
When I'm not at a desk, I'm at home with Joya and seven cats. I know how that sounds. It's also true. The cats have personalities, opinions, and a strong stance on the kitchen counter. Joya, mercifully, does not ask for breakfast at 4 a.m.
And one co-founder
Joya
Co-founder · MD, NextBangla Ltd.
- MishtiSenior nap officer
- PakkaChief mischief officer
- NuriWindow watcher
- KaluLap warmer
- BobaTreat receiver
- PoriLate-night philosopher
- JhilikCurtain destroyer
Ventures that didn't work.
Not every venture made it. There were several — a marketplace that ran out of runway in 2014, an IT-staffing arm that we shut down rather than dilute the brand, a product that found product-market fit two years after we needed it to. Each one cost real money, real time, and real relationships. None of them were clean failures. All of them taught me something I'm still using.
- Distribution beats product. The 2014 marketplace had the best UX of anything we shipped that year. Nobody could find it.
- Hiring before revenue is a tax. Twice we scaled headcount on the assumption deals would close. They closed slower. We carried the cost.
- Pivot fast or shut down faster. Two of the failures dragged because we kept “improving” instead of accepting they wouldn't work.
Voxly + 9 more.
Voxly is a real-time voice-to-text product for Bangla and other under-served South Asian languages. It's the one I'm shipping fastest. The other 9 are at various stages — three in beta, four still building, two figuring out if they should exist at all.
Operating at AI velocity.
Building 10 AI startups in parallel with a small team is teaching me things in real time about model selection, eval pipelines, cost-per-conversation economics, and the difference between “AI feature” and “AI product.” I'll be writing about it on the blog.
If any of this resonated, let's talk.
I'd rather hear from a founder who's actually building than someone trying to sell me on AI. If that's you, the door is open.