I turn manual business workflows into interactive, reliable, and maintainable web products.
- 🛠️ I build production applications across React, TypeScript, Node.js, and Hono, from requirements and API design through deployment.
- ✨ I care about accessible interfaces, thoughtful motion, and the tiny interactions that make software feel effortless.
- ⚙️ I enjoy workflow automation, API integrations, internal tools, and eliminating repetitive work.
- 🎓 Former Microsoft Learn Student Ambassador and SDE intern, now delivering full-stack features in production.
- 🌍 Based in Faisalabad, Pakistan; open to remote roles, freelance work, and open-source collaboration.
| Challenge | What I changed | Result |
|---|---|---|
| Slow third-party synchronization | Batched and redesigned Jobber API requests | Reduced sync time by ~70%, from 80 minutes to 20–35 minutes |
| Manual commission operations | Built a Jobber-integrated commission and worklog platform | Replaced spreadsheet processing and supported 20–30 weekly submissions |
| Fragile client-side data fetching | Migrated application data fetching to TanStack Query | Eliminated race conditions and unnecessary re-renders |
| Repetitive management workflows | Developed custom workflow automation systems | Reduced required management effort by 50% |
| Area | Technologies |
|---|---|
| Front end | React · Next.js · Vue.js · TypeScript · TanStack Query · Tailwind CSS · Framer Motion · shadcn/ui · Storybook · Vite |
| Back end | Node.js · Express · Hono · Prisma · REST · GraphQL |
| Data | PostgreSQL · Redis · SQL · NoSQL · third-party API integrations |
| AI / ML | Python · TensorFlow · Keras · TensorFlow.js |
| Delivery & tools | Cloudflare Workers · GitHub Actions · Docker · Linux/VPS · DigitalOcean · pnpm · Figma |
A multi-tenant production application for employee worklogs and commission management. It uses verified CRM data as its source of truth and supports configurable commission rules, approvals, payments, refunds, and duplicate prevention.
TypeScript · React · Tailwind CSS · Hono · Docker · CI/CD · DigitalOcean
Explore the live application →
A real-time sketch classifier powered by a CNN trained on Google's QuickDraw dataset. The model is exported to TensorFlow.js and runs entirely in the browser—no server round trip required.
Python · Keras · TensorFlow · TensorFlow.js · Vue.js
More work and case studies are available at jahanshah.dev.
- Building and maintaining production workflow-automation software
- Growing deeper in backend architecture, microservices, and cloud deployment
- Looking for meaningful engineering problems and people who care about solving them well