Waste Flow turns industrial by-products into nearby supply. Inspired by plant tours where valuable streams like paper sludge, kiln dust, low-grade heat, and CO₂ are landfilled while neighbors buy the same inputs, we built a FastAPI MVP that ingests facility waste and need profiles, normalizes compositions, ranks matches with a weighted Jaccard and Haversine blend, and uses Gemini to generate concise, risk-aware briefs with concrete next steps. We shipped a runnable API with bulk load, match, and explain endpoints plus a Texas seed dataset to demo cross-industry wins such as pulp to cement and refinery heat or CO₂ to greenhouses. Challenges included messy materials data, balancing chemistry with logistics, prompt grounding, and merge conflicts that pushed us to tighten our Git workflow with feature branches, smaller pull requests, and fast reviews. We are proud of a clear and tunable scoring model, instant demoability, and practical AI-assisted summaries, and we learned that constrained prompts and small UX lifts like company, state, and city autocomplete dramatically speed adoption.

Built With

  • backend:
  • backend:-fastapi
  • frontend:
  • git/github
  • icons
  • lucide
  • lucide-icons-|-backend:-fastapi
  • next.js-15.2.4
  • npm
  • pydantic-|-dev-tools:-vs-code
  • python-3.9
  • radix-ui
  • react-19
  • shadcn/ui
  • tailwind-css
  • typescript
  • uvicorn
Share this project:

Updates