What I build,
and how fast.
Five services, one method: AI does the mechanical work, I do the engineering. You get production software on a timeline that used to require a team — without the codebase you'd have to apologise for later.
Five services, one method: AI does the mechanical work, I do the engineering.
Where the weeks actually go.
Scope with AI, decide with judgment
01/04I pressure-test the spec against edge cases before a line is written, so week one builds the right thing instead of discovering it in week three.
Generate the boring 60%
02/04Scaffolding, CRUD, types, test fixtures, migrations — the parts that are mechanical get generated in minutes, not days.
Spend the saved time on the hard 40%
03/04Architecture, data modelling, performance and the interaction details are where I actually spend the budget you saved.
Review everything, ship small
04/04AI-assisted review plus my own read on every PR, with tests and a preview deploy on each push. Nothing lands unread.
What AI doesn't change: someone still has to own the architecture, catch the subtle bug, and say no to the feature that will sink the timeline. That part is still me — the tooling just means you pay for judgment instead of typing.
Five ways I can help.
I start from a setup I already trust rather than picking tools from scratch each time. See what I build with →
AI Web Development
Full web applications built with an AI-accelerated workflow — and, when the product calls for it, AI features built into the app itself: chat, semantic search, extraction and automation. Same engineering standards, a fraction of the calendar time.
8 capabilities+
- Next.js & React application builds
- AI chat, assistants & agents
- Semantic search / RAG over your content
- Document & data extraction pipelines
- OpenAI & Claude API integration
- AI-assisted code review and testing
- Streaming UIs and token-cost budgeting
- Typed, tested, human-reviewed code
MVP Engineering
For founders who need something real in front of users before the next raise, the next board meeting, or the end of the runway. I scope hard, build only what proves the thesis, and ship it instrumented so the next decision comes from data instead of opinions.
8 capabilities+
- Scoping and ruthless feature triage
- Auth, billing and onboarding out of the box
- Database design that survives v2
- Admin dashboard and internal tooling
- Analytics & event instrumentation
- Deploy, domains and monitoring
- Weekly demo builds you can click
- Clean handoff — or I keep building
Shopify & CRO
Custom Shopify themes and headless Hydrogen storefronts, plus the conversion rate optimization that turns the traffic you already have into orders. Build and optimize are one service here, because shipping a fast store that nobody buys from is only half a job.
8 capabilities+
- Custom theme & Hydrogen storefronts
- Figma-to-Shopify development
- Speed & Core Web Vitals work
- Conversion audits & funnel analysis
- A/B testing to significance
- Cart & checkout optimization
- Custom sections and app integrations
- Ongoing support and iteration
AI Booking Systems
A conversational booking system for clinics, salons and trades: customers chat on WhatsApp or your site, the assistant checks real availability and books the slot. The scheduler owns scheduling correctness; the model only handles the conversation, and it never writes to your database directly.
8 capabilities+
- WhatsApp & on-site chat booking
- Real availability via Cal.com
- Confirmations and reminders
- Owner alerts on every booking
- Admin view to edit or cancel
- Guardrails and a human fallback
- Multi-language (English & Bangla)
- Live in about 3 weeks
Full-Stack & APIs
Node and Express APIs, MongoDB data modelling, authentication, background jobs and third-party integrations — built typed end to end so the front end and the back end never disagree about what a user is.
8 capabilities+
- REST & typed API design
- MongoDB schema and index design
- Auth, roles and permissions
- Payments & third-party integrations
- Background jobs and queues
- React Native mobile clients
- Test coverage on the critical paths
- Docs a new dev can actually follow
Three ways to start.
Fixed-scope sprint
01/03One clear outcome — a feature, an integration, an audit — scoped, quoted and delivered in one to two weeks.
Product build
02/03An MVP or a full application, from schema to launch. Weekly demo builds, a fixed scope and a real launch date.
Ongoing partner
03/03A reserved slice of my week for teams that keep shipping — features, fixes, experiments and the odd fire.
Built
for speed.
Six steps from first message to production. No bloat, no ceremony — just momentum and clean commits.
You reach out
A role spec or a few lines about your project. No long brief required — I'll ask the right questions.
We get on a call
A quick call to align on scope, stack and timeline — or to talk through the role and team.
I scope the work
Clear deliverables, a realistic timeline and a tech plan you can reason about. No surprises later.
I build & commit
Small PRs, clear commit messages and a preview deploy on every push, so you always see progress.
We review together
Real feedback on a real deploy. We refine, test and tighten until it's right, not just done.
We ship it
Production deploy, docs and a clean handoff — with the support to keep moving after launch.
The questions
I usually get.
Which service do I actually need? +
If you have an idea and no product, that's MVP Engineering. If you have a product and a backlog, that's AI Web Development. If you have a store and traffic that isn't converting, that's Shopify & CRO. Tell me the situation and I'll say which one — including if the answer is none of them.
How do you price this? +
Fixed price against a fixed scope for sprints and product builds, and a monthly rate for ongoing work. You get the number before anything starts, and it does not move unless the scope does.
Can you add AI features to an existing app? +
Yes — assistants, semantic search over your own content, extraction and workflow automation, wired into what you already have with cost and latency budgets agreed up front.
What if the AI-generated code is wrong? +
Then it gets rewritten before you ever see it. Generated code is a first draft, not a deliverable — everything goes through tests, review and my own read before it reaches your repo.
Do I own the code? +
Entirely. Your repo, your accounts, your infrastructure from day one — plus docs and a handoff walkthrough at the end.
How fast can you start? +
Usually within a week or two, sometimes sooner for small sprints. Send the details and I will tell you the honest next opening.