Contact person: AI station

Phone:Show

Email:Show

Location: Goreakothi, India

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"I’ve already built a full-stack AI platform that blends content creation, image generation, and productivity tools. It runs on a Node.js + [login to view URL] API backed by PostgreSQL (Neon) and a React / Vite / Tailwind front-end, with Clerk handling authentication for both free and premium tiers. Core services are live on Vercel with CI/CD through GitHub Actions and Docker, and I’ve wired in Gemini AI, OpenAI-compatible endpoints, and ClipDrop.

Now I want to double-down on the part users love most: intelligent image editing. The immediate goal is to refine and extend this pillar so the experience feels as magical as the underlying models.

Scope of work
• Intelligent image editing – background removal must be pixel-perfect on web and mobile resolutions, and I want to add creation workflows that let users generate brand-new visuals (AI-generated images) or start from designer-ready templates (template-based image creation).
• Keep latency low; the current Express routes should respond in under 2 s for 4 K images. If necessary, optimise with worker queues or serverless functions.
• Preserve secure role-based access with Clerk so premium limits and usage quotas remain enforced.
• Extend the community gallery so newly created or edited images can be liked, remixed, and attributed to their creators without exposing private originals.
• Ship through the existing CI/CD pipeline; Docker-files and GitHub Actions workflows must stay green.

Tech context
Node.js 20, Express 4, PostgreSQL + Neon, Prisma ORM, React 18, Vite, Tailwind CSS, shadcn/ui and Radix primitives. Deploy target is Vercel; image processing can run on Vercel Functions, Edge, or a separate GPU-enabled service if you have a strong preference.

Acceptance
A merge-ready PR demonstrating:
1. Background removal that beats the current ClipDrop baseline on edge cases I’ll provide.
2. An end-to-end image creation flow covering both AI-generated and template-based options, saving results to the gallery with proper access rules.
3. Updated tests and docs so another dev can spin up the stack with `pnpm dev` and see everything working.

If you have direct experience wrangling large images in a serverless environment and can speak both design language and low-level optimisation, I’d love to see what you can do." (client-provided description)