Full-Stack Engineer, web and mobile. For 4+ years I have built booking, payment and learning products for clients in Australia, Sweden and the UK.
I start with what can go wrong before I build a feature. The case studies show the problem, the decision and the trade-off: https://www.mehedisarkar.com/work?ref=github
- Booking platforms. A 10-minute slot hold, so two people cannot book the same court at the same moment. Reminders, cancellations and club payouts. Case study
- Payments. Stripe across 4 products: manual capture, Connect payouts, refunds and 3D Secure. Money only moves when the booking is really going ahead. Case study
- Inherited systems. On one salon booking system I documented 31 defects, fixed them in priority order and cut the mobile payload from 17.2 MB to 1.7 MB. Case study
- AI content pipelines. Gemini OCR that turns scanned textbooks into typed content blocks, with per-page timeouts, retries and recovery after a worker crash. Case study
- Mobile apps. Led a mobile team of 3 developers on 2 products and owned the App Store and Google Play releases. Worked on a padel booking app with 100K+ Google Play downloads.
TypeScript, Node.js, Express, NestJS, React, Next.js, React Native, Expo, MongoDB, MySQL, Redis, BullMQ, Stripe, Docker, GitHub Actions, Vercel, AWS.
- desco-balance-check: Telegram bot for prepaid electricity customers in Dhaka.
- task-flow: project management app with a permission-aware AI assistant.
- MindPop: offline-first brain-break games app, live on Google Play.



