AI Product Development Services

I'm a freelance AI product developer from Rajshahi, Bangladesh, working under the name Mursalin's Desk. I build AI agent SaaS products, AI-powered mobile apps with React Native and Expo, MVPs for startups, AI integrations for existing products, and full-stack web platforms on Next.js, NestJS and PostgreSQL. I also take on AI strategy and fractional CTO engagements for founders who need senior technical leadership without a full-time hire.

I'm based in Rajshahi, Bangladesh, and also work from Dhaka. Most of my clients are remote — startups and small companies in the US, UK, Canada, Australia and the EU. I can meet in person in Rajshahi or Dhaka when that helps, but every project I run is set up to work fully remotely from day one: shared repo, project board, async updates and scheduled calls.

Yes. You can hire me directly, without an agency or marketplace in between. We agree a scope and price, sign a contract with IP assignment and an NDA if you need one, and you pay by milestone through Wise, Payoneer or bank transfer. I have shipped production LLM agents, RAG systems and AI mobile apps for overseas clients this way, with code in your repository and infrastructure in your accounts.

I work in UTC+6 (Asia/Dhaka). That is 5 hours ahead of London and 10-11 hours ahead of New York, so I overlap with the full UK and EU afternoon and with US mornings and evenings. I keep a fixed daily window for calls, answer messages within a working day, and run the rest of the project asynchronously with written updates, so you never wait on my clock to make progress.

Book a free 30-minute call or send a short brief through the contact form. On the call I ask about the problem, users, budget and deadline, and tell you honestly whether one of my packages fits or a custom scope is better. Within two working days you get a written proposal with scope, milestones, price and timeline. Once you approve it and pay the first milestone, work starts the following week.

I invoice in USD and accept Wise, Payoneer and international bank transfer; Stripe invoicing is available for card payments. Fixed-price packages are split into milestones: typically 40% to start, 40% at the mid-point demo, and 20% on handoff. Hourly work is tracked and invoiced weekly, and retainers are billed at the start of each month. You always receive a proper invoice you can expense.

A single-purpose AI agent SaaS with sign-in, Stripe billing and usage metering starts at $1,499 and takes 2-3 weeks. A multi-tenant product with RAG, agent workflows, team workspaces and observability is $4,999 over 5-8 weeks. Enterprise multi-agent platforms with model routing, SSO and eval pipelines start at $12,999. US agencies quote $20,000-50,000 for comparable agent builds, so you get senior work at offshore rates.

My MVP packages run from $999 for one core workflow with auth and a live URL (2-3 weeks), to $2,999 for a launch-ready product with Stripe payments, an AI feature and an admin dashboard (4-6 weeks), to $7,499 for a pilot-grade build with SSO and integrations. Local agencies in Bangladesh typically quote $6,000-10,000 for a lean MVP, and US shops $25,000 and up, so the difference is real without cutting scope.

Hourly work is $100-150 per hour depending on scope and stack, tracked weekly with a two-hour minimum. Monthly retainers with reserved hours start at $500. For context, senior AI developers in the US bill $150-250 per hour and UK contractors a similar range, while typical Bangladesh outsourcing rates are $15-35 per hour for mid-level work. I sit between those: senior product engineering at a rate that still leaves room in a startup budget.

Both. Fixed-price packages suit projects with a clear scope: you know the total, the milestones and the delivery date before we start. Hourly billing suits evolving work — bug fixes, audits, AI experiments, or joining an existing team. Retainers cover ongoing maintenance and roadmap work with a reserved weekly allocation. Every proposal states which model applies and what is excluded, so there are no surprises on the invoice.

Your own accounts carry the running costs: hosting on Vercel, Render or AWS is usually $20-100 per month for an early product, a managed Postgres like Supabase or Neon is free to $25, and LLM API fees depend on usage — a few dollars a month for a small tool, $50-500 for a growing SaaS. Every AI build includes cost caps and a usage dashboard, and I give you a running-cost estimate in the proposal.