B2B Produce Trading Platform, from Zero to Operation
A large online produce trading platform, built from scratch and operated, ending in a sale of the business.



Challenge
B2B produce trading among large agricultural producers, wholesalers, and retailers had long relied on phone, fax, and offline processes. It was inefficient and opaque, and far less digital than the volume of trade called for. Wholesalers calculated profit and loss by compiling figures by hand every week, and an aging desktop database application was maintained by a single developer. If that person left, no one would dare touch the system.
Solution
Scope of work: Building the trading platform from scratch through to operation, plus supporting systems along the distribution chain (orders, logistics, profit and loss, forecasting) and project team management.
- Trading platform from zero to one: Built and operated a large B2B online produce trading platform from scratch on FastAPI and PostgreSQL, delivered end to end from requirements, design, and development through testing and operation, with the back-office admin system built alongside.
- Distribution chain systems: A web-based replacement for the legacy order system (removing the single-developer dependency and creating a structured data foundation), a logistics picking system, and price forecasting and purchasing optimization.
- Profit and loss automation: A P&L automation system for a large fruit and vegetable wholesaler (FastAPI, Plotly Dash) fully automated the weekly manual process, saving 20 or more hours a month and enabling daily P&L analysis.
- AI and data in the field: Projects such as produce grade recognition AI and price forecasting brought data analysis and AI to farms, logistics, and retail operations.
- Team and collaboration: Built and managed a project team of up to 30 people, set up a long-term collaboration with an overseas engineering team, and took part in a smart agriculture demonstration project.
Results
- Platform built and in operation: Built from zero to one and operated on an ongoing basis.
- Team of up to 30 people: The project team was assembled and managed.
- Startup exit: The business was eventually sold. This project was also the start of the long-term collaboration with overseas engineers, which continues today.
See other case studies
Eight platform-wide sales events in over two years with zero major incidents, load tested at up to 8 times normal traffic.
Video data and transaction data connected across about 500 stores, cutting the investigation of a suspicious transaction from a week to a day.