Mai Shan Yun Inventory Intelligence

Inspiration

In restaurant management, balancing ingredient supply and demand can feel like walking a tightrope — too much inventory leads to waste, too little leads to shortages. We were inspired by Mai Shan Yun (Wheat · Mountain · Cloud) to represent the perfect balance between nature, precision, and flow. Our goal: build a beautiful, data-driven system that helps restaurants think ahead — forecasting needs, reducing waste, and transforming raw spreadsheets into clear, actionable insight.

What it does

Mai Shan Yun Inventory Intelligence is an interactive dashboard that empowers restaurant managers to:

Track inventory levels, purchases, usage, and shipments in real time.

Identify ingredients that are running low or overstocked.

Forecast future ingredient demand using PyTorch-based predictive models.

Visualize relationships between menu item sales and ingredient consumption.

Monitor shipment frequency and delays through dynamic charts.

Toggle between English and Chinese interfaces with light/dark modes.

Securely sign in via Google OAuth, ensuring safe and personalized access.

The result is an intelligent platform that blends AI analytics, visual storytelling, and operational control — helping kitchens run smoother than a perfect stir-fry.

How we built it

Frontend: Built in Next.js + TypeScript, styled with Tailwind CSS and ShadCN/UI, animated with Framer Motion.

Backend: Powered by FastAPI to handle data ingestion, processing, and ML inference.

Data: Parsed and merged .csv and .xlsx datasets (purchases, shipments, sales, ingredient usage) via Pandas.

Forecasting: Implemented a PyTorch time-series model (LSTM/TCN) to predict ingredient demand.

Authentication: Integrated NextAuth.js with Google Sign-In.

Storage: Managed cleaned and structured data using PostgreSQL.

Visualization: Designed rich interactive charts with Recharts and Chart.js.

Deployment: Containerized via Docker; frontend hosted on Vercel, backend on Render.

Design: Inspired by Chinese color palettes — crimson, jade, and gold — symbolizing wheat, mountain, and cloud.

Challenges we ran into

Dataset integration: Merging messy purchase, shipment, and sales data required careful cleaning and schema alignment.

Forecasting accuracy: Balancing model complexity and real-time inference speed was tricky.

Localization: Building a seamless bilingual (English/Chinese) interface required managing parallel translations and RTL spacing.

UI performance: Keeping complex, animated dashboards smooth while handling large datasets took optimization and caching work.

Authentication flow: Getting Google OAuth and NextAuth to cooperate locally with environment variables was a learning curve.

Accomplishments that we're proud of

Created a fully functional, visually immersive dashboard with a cohesive theme.

Successfully integrated AI forecasting that gives actionable reorder predictions.

Designed a modern bilingual interface that adapts to light and dark modes.

Achieved a fluid, flame-animated Chinese-inspired UI without sacrificing performance.

Developed clear, actionable insights — turning restaurant chaos into clarity.

What we learned

How to bridge data science and UI design into one cohesive experience.

Practical data cleaning and forecasting techniques using Pandas and PyTorch.

The power of storytelling through visualization — how good visuals can make analytics intuitive.

Real-world challenges of handling multi-source restaurant data and time-series prediction.

How thoughtful UX decisions (language, theme, animation) elevate user trust and engagement.

What's next for Mai Shan Yun Inventory Intelligence

Integrate real-time POS (Point-of-Sale) data for continuous updates.

Add automated purchase order generation when stock forecasts drop below thresholds.

Deploy forecasting to a cloud ML API for scalable performance.

Enhance cost-optimization analytics with vendor comparison insights.

Expand beyond restaurants — adapting Mai Shan Yun for cafés, catering services, and food suppliers.

Continue refining the visual identity to fully capture the meaning of Wheat · Mountain · Cloud — grounded in data, elevated by insight, and powered by the flow of innovation.