About · AI-assisted developer & senior Bubble.io engineer

4,000+ hours delivered
100% job success on Upwork
Top Rated Plus Upwork status
12yr BD + SDE

Most software ships late and broken for the same reason: nobody really understood what was being built before the building started. I spent close to a decade watching that pattern from the business side, sitting between clients and developers. In 2022 I crossed over to building. The pattern still happens. It just costs less when I’m in both chairs.

That’s why the emails I get all sound alike. They come from founders, operators, and in-house teams. The story is always the same: their developer reads the brief and builds the wrong thing. They read the technical answer and approve the wrong thing. The project drifts in the gap. They hire me because I’ve sat in both chairs, and because responsible development is the part that doesn’t change when the stack does.

That’s what I do. Seven recent case studies show it across compliance software, AI products, and marketplaces. The stack varies. The discipline doesn’t.

How I work, not what I work on.

The tech stack is a means. The discipline around it is what determines whether the project ships, scales, and survives the next two iterations. Three principles I work from:

Discuss in detail, then write it all down. Specs, blockers, scope changes, dead ends. Each gets a doc, a paragraph, or a file. AI-assisted work depends on that context to stay useful, and a project six months in depends on it more. The conversations happen out loud. The history stays on disk.

Architecture changes the direction of everything that comes after. So I think it through, audit twice, and walk the client through the trade-offs in plain English before any code goes down. What we’re choosing, what we’re giving up, what it costs six months out. The decision happens once, with the client in the loop.

Security, data integrity, reusability, and scalability stay on the list regardless of scope. The MVP gets the same row-level security as the multi-tenant SaaS. The internal tool gets the same input validation as the public product. Because the project that works is the one that grows.

The twelve-year arc

The practice today, and the twelve years behind it.

2026

Now

Claude Code, Cursor, Lovable, Bubble.io, n8n, Supabase, Vercel: the mix that fits the client, not the other way around. Currently shipping GoBD V3 and ongoing engagements.
2025

Shipped multi-tenant V2 + responsive-engine migration

GoBD V2 turned the MVP into a partner-distributable white-label SaaS with Bubble.io sub-apps, PDF Monkey, n8n, and AI integration for compliance language. ShareTalent migration to Bubble's new responsive engine completed (17 months as lead dev, zero data loss).
2024

Reached Top Rated Plus

Earned Top Rated Plus on Upwork. First full calendar year of solo AI + Bubble.io work: Construction Market MVP in three sprints, RealClear (six AI marketing tools, via Zeroic), and GoBD V1 admin platform for a German compliance SaaS. Took on ShareTalent's migration to Bubble's new responsive engine as lead dev.
2023

Started freelancing on Upwork

Went independent in June. Earned Rising Talent, 100% Job Success, and Top Rated badges within the year. Every engagement since mid-2023 has been mine end to end: senior Bubble.io dev, solo on the work. Shipped FoodIWant (Stripe rescue + marketplace expansion to six verticals).
2022

Crossed over to building

Joined Infex Biztech as a Bubble.io dev. Shipped Sikrits (live coaching + quiz portal on a course platform) and the HR Management Portal (multi-tenant Bubble.io SaaS, 120k+ row reporting).
2014

Started in business development

Ahmedabad. Business development on software projects. Drafted requirements, sat across the table with developers and clients, watched projects ship and fail. The pattern was always the same: most failures came down to how well the developer understood the business requirement, and how cleanly that got translated into a technical one.