JAINIL DESAI
Full-Stack Engineer · Ahmedabad, India
Available for work
Perceive. Build. Ship.
Scroll to explore
I'm a full-stack engineer drawn to AI, ML and automation — from real-time computer-vision models to a graph-based recommendation engine that lifted revenue 30% across 2,000+ stores. Fast systems under a UI that feels effortless.
Plantation & CO₂ Platform
A platform to manage large-scale tree plantations and prove their impact. Every tree gets a QR identity; field officers geotag and update growth from mobile. It tracks survival, computes CO₂ absorption, gives donors a dashboard with photos and certificates, and an AI assistant turns plantation data into instant impact reports.
ASL Sign Detector
Real-time American Sign Language recognition — webcam hand-tracking via MediaPipe feeds a TensorFlow model that predicts letters, then builds words with smoothing and fuzzy spell-correction.
LeetHub Sync
A zero-touch pipeline that auto-commits LeetCode solutions to GitHub with structured messages and updates a Notion progress board via GitHub Actions — no manual tracking, ever.
AI / ML
01- Computer vision (MediaPipe)
- Recommendation systems
- Sentiment analysis
- Gemini API
Automation
02- CI/CD pipelines
- GitHub Actions
- Webhooks
- Workflow tooling
Full-stack web
03- React / Next.js
- Node / Express
- REST & GraphQL
- TypeScript
Data & Cloud
04- MongoDB · PostgreSQL
- Supabase
- Vercel / VPS
- Docker
An AI-native workflow. Claude Code does the heavy lifting; I steer the architecture, systems and deploys.
Full-stack web apps
End-to-end products — Next.js front-ends, Node/Express APIs, auth, payments and a clean, fast UI.
AI & automation
LLM features, recommendation and computer-vision systems, and pipelines that remove manual work.
Backend & infrastructure
APIs, databases, VPS and cloud setup, domains and deployments that stay fast and reliable.
Software Developer — Webrex Studio
- Drove a +30% increase in merchant revenue across 2,000+ stores by designing a graph-based product recommendation engine (Amazon-style co-purchase, edge weights updated from order co-occurrence) ranking related products at low latency.
- Lifted installs 80% and raised activation from 30% → 80% by rebuilding the embedded merchant admin and storefront logic from scratch in clean React.
- Improved app performance 40% by refactoring storefront rendering paths and tightening Admin GraphQL queries.
Freelance Full-Stack Developer — Self-employed
- Delivered 4 paid engagements across e-commerce and SaaS — owning requirements → architecture → deployment → handover.
- Built an end-to-end store (catalog, cart, checkout, admin) on Next.js + Tailwind with Supabase auth & Postgres, deployed to VPS and Vercel.
- Cut documentation effort ~70% with a workflow automation tool generating structured summaries from raw inputs.
Frontend Developer Intern — Digital Dose
- Reduced frontend load time 25% by engineering an Express/Node REST API over MySQL with query batching and response-shape optimization.
- Integrated Razorpay end-to-end (init → callback → reconciliation) with idempotent webhook handling.
- Cut a client workflow from 2 min to under 30 s by automating manual steps with a Node/Python pipeline.