Selected Work
Selected Work

Multi-tenant Sticker Commerce & AI Ops Toolkit

Worked on a multi-tenant sticker commerce system together with AI image generation, OCR quality inspection, payments, SEO and Cloudflare deployment workflows.

Shopify APIPythonPaddleOCRPayPalStripeCloudflare WorkersR2
Sep. 2025 - Mar. 2026
Shenzhen
Shenzhen Meiyunjing Trade Co., Ltd.
RoleAI Full-Stack Development Intern

Project Context

This project was not a single feature, but a combination of business systems and supporting tools around customized sticker sales.

The team needed a system that could support multiple tenants, customer-facing ordering, AI-assisted asset generation, production checks, payment handling, and a deployable online workflow. That meant the engineering work had to connect front-end business flow, operational tools, and infrastructure delivery.

My Responsibilities

I participated across several layers of the project instead of staying in only one role.

  • Supported the multi-tenant commerce workflow around sticker customization.
  • Worked on AI image generation related process integration.
  • Participated in OCR-based text inspection for production quality control.
  • Helped with payment access, SEO optimization and Cloudflare-based deployment workflows.

Business and Technical Flow

The system covered the full path from user demand to operational execution:

  1. Customers configure or upload custom sticker requirements.
  2. The system manages assets and order-related data for different tenants.
  3. AI generation and processing tools help prepare visual materials.
  4. OCR inspection helps identify text issues before production.
  5. Payment flows and storefront delivery complete the customer side.
  6. Cloudflare services support deployment and asset delivery.

Because these parts were coupled in the real workflow, the engineering work also had to stay practical and integrated.

Key Problems Solved

Multi-system coordination

The hardest part was not one algorithm or one interface. It was getting storefront logic, AI tools, OCR checks and deployment workflows to work together without breaking the business process.

Production-oriented quality checks

For a customization business, output quality matters. OCR inspection helped reduce manual review pressure and added a repeatable check before assets entered production.

Delivery and online operations

Engineering work only becomes valuable when it can actually support operations. SEO and deployment work mattered because they directly affected whether the store could be found, maintained and updated smoothly.

Outcome

This project expanded my understanding of full-stack AI application delivery in a business context. I was able to work on the connection points between commerce, AI tools, operational inspection and deployment, which is very different from building an isolated demo.

It also helped me understand that real-world AI projects often succeed because the surrounding business workflow is carefully stitched together.

Reflection

What stayed with me most is the need to think in systems. In a commercial workflow, each module may look independent, but users experience the entire chain. The engineer's job is to reduce friction across that whole chain, not only optimize one part in isolation.