Back to Portfolio

StockMind

React 19FastAPIPostgreSQLGoogle Gemini API

Overview

Full-stack AI inventory platform. Includes an MCP-assisted EC2 remote testing workflow. AI drafts purchase orders (human review required).

Links & Evidence

View on GitHub

Problem & Contribution

Problem: SMB inventory systems often lack automated reordering intelligence, requiring manual tracking of stock levels.

My Contribution: Developed this project with substantial AI assistance. I defined the data models, guided the implementation, and configured the services.

Architecture & Implementation

  • Frontend: React SPA initialized with Vite, styled with Tailwind CSS, utilizing Zustand for state.
  • Backend: FastAPI REST API using asynchronous sessions (SQLAlchemy 2.0 with asyncpg) to PostgreSQL.
  • AI Integration: Google Gemini API autonomously drafts structured Purchase Orders based on reorder thresholds (human review is explicitly required before fulfillment).

Testing & Limitations

Testing Workflow: Conducted deployment tests via an MCP-assisted EC2 remote workflow.

Limitations: This project is a proof-of-concept exploring AI integration. The remote testing established the deployment workflow, but does not constitute a fully validated, production-ready system. AI-generated purchase orders must always be reviewed by a human.