Setting up Python backend:

1. Make sure you have Python3.11 installed
2. python3.11 -m venv .venv
3. source .venv/bin/activate
4. pip install -r requirements.txt
5. To run: uvicorn app.main:app --reload

Installing .env's so api's work

1. cp .env.example .env
2. npm install
3. npm run dev

Goal: Establish a working development environment.

• Initialize GitHub repo
• Create frontend (Next.js recommended)
• Create backend (FastAPI recommended)
• Add one sample CSV to backend folder
• Hardcode one supplier JSON (for testing)
• Verify frontend ↔ backend connection with a test API call

Goal: Load stock CSVs and compute essential supplier metrics.

Tasks:

• Load all supplier CSVs (CLF, X, STLD, CMC, MT, FRD)
• For each supplier, compute:
  • Average closing price (cost proxy)
  • Daily returns
  • 30-day volatility (risk proxy)
  • Trend slope (optional)
  • Average trading volume (optional)
• Store results in an in-memory dictionary
• Create a supplier metadata JSON containing:
  • Plant location (lat/lon)
  • Steel process (EAF vs BF)
  • Carbon intensity (CO2/ton)
  • Default logistics mode (truck/rail/ship)

Deliverable: Backend can compute cost_score and risk_score.

Goal: Implement optimization math and scoring system. Tasks:

• Implement Haversine distance calculation
• Add transport emissions model:
  • Emission factors for truck/rail/ship
  • Distance × tonnage × emission factor
• Compute material emissions using CO2 per ton
• Combine material + transport emissions into a carbon score
• Compute final composite score:
• final_score = w_cost * cost_score + w_co2 * co2_score + w_risk * stability_score + w_logistics * logistics_score
• Build /optimize API endpoint:
  • Inputs: location, tonnage, weights, carbon price
  • Output: ranked list of suppliers + scores + distances + emissions

Deliverable: Backend can rank suppliers using cost, carbon, risk, and logistics.

Goal: Allow users to enter inputs and visualize ranked suppliers. Tasks:

• Build the main input form:
  • Plant location
  • Tonnage needed per year
  • Slider: Cost ↔ Carbon
  • Carbon price input
  • "Optimize Supplier" button
• On submit, call backend /optimize
• Render results in a table:
  • Supplier name
  • Final score
  • Total CO2
  • Effective cost
  • Transport distance
• Optional: Add charts (bar, radar, line)

Deliverable: Frontend connected to backend with dynamic results table.

Goal: Display suppliers and shipping routes on an interactive map. Tasks:

• Use Leaflet.js or Mapbox
• Plot supplier plant locations
• Plot user’s plant location
• Draw route lines
• On hover: show distance and transport CO2

Deliverable: Interactive map showing supply routes.

Goal: Automatically explain why the top supplier was recommended. Tasks:

• Feed optimization result into an LLM
• Generate short text explanation summarizing cost, carbon, and logistics reasoning

Deliverable: AI-generated summary on results page.

PHASE 6 — Polish and Pitch (≈ 1–2 hours) Goal: Refine UI/UX and prepare final presentation. Tasks:

• Add simple styling (Tailwind recommended)
• Create a "How It Works" page
• Prepare pitch script:
  • Problem → Solution → Demo → Impact → Future Work

Deliverable: Polished, demo-ready project.