Use pandas in a web backend when:
☐ Users upload CSV or Excel files
☐ You need grouped summaries (per month, per category)
☐ You export reports for download
☐ Data needs cleaning before saving

Avoid pandas when:
☐ You only read or write single database rows
☐ The database can do the aggregation with SQL efficiently
☐ Data is too large for memory (use SQL or chunking)

pip install pandas openpyxl fastapi uvicorn python-multipart
# openpyxl: Excel read/write
# python-multipart: file uploads in FastAPI

import pandas as pd
from fastapi import FastAPI, UploadFile

app = FastAPI()
REQUIRED = {"name", "email", "amount"}

@app.post("/import")
async def import_csv(file: UploadFile):
    df = pd.read_csv(file.file)
    missing = REQUIRED - set(df.columns)
    if missing:
        return {"ok": False, "missing_columns": sorted(missing)}
    df["amount"] = pd.to_numeric(df["amount"], errors="coerce")
    bad = df[df["amount"].isna() | df["email"].isna()]
    good = df.drop(bad.index)
    # save good.to_dict("records") to your database here
    return {"ok": True, "imported": len(good),
            "bad_rows": (bad.index + 2).tolist()}  # +2 = header + 1-based

curl -F "file=@payments.csv" http://127.0.0.1:8000/import
# {"ok":true,"imported":48,"bad_rows":[7,19]}

import pandas as pd
from fastapi import FastAPI

app = FastAPI()

def load_orders():
    # replace with a database query, e.g. pd.read_sql(...)
    return pd.DataFrame({
        "date": pd.to_datetime(["2026-01-05", "2026-01-20", "2026-02-03"]),
        "amount": [1200, 800, 1500],
    })

@app.get("/reports/monthly")
def monthly():
    df = load_orders()
    out = df.groupby(df["date"].dt.strftime("%Y-%m"))["amount"].sum()
    return [{"month": m, "total": int(t)} for m, t in out.items()]

from io import BytesIO
from fastapi.responses import StreamingResponse

@app.get("/reports/monthly.xlsx")
def monthly_excel():
    df = load_orders()
    buf = BytesIO()
    df.to_excel(buf, index=False, sheet_name="orders")
    buf.seek(0)
    return StreamingResponse(
        buf,
        media_type="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
        headers={"Content-Disposition": "attachment; filename=orders.xlsx"},
    )

Project: Small Business Expense Tracker
Frontend: React or Django templates
Backend: FastAPI or Django
Database: PostgreSQL or SQLite
pandas features:
  ☐ Bulk CSV import with row-level errors
  ☐ Monthly and category summaries API
  ☐ Excel export of filtered data
Frontend features:
  ☐ Upload page
  ☐ Dashboard with two charts
  ☐ Download button

Endpoints
POST /import              → upload CSV, returns imported count + bad rows
GET  /expenses            → list with filters (date, category)
GET  /reports/monthly     → [{month, total}]
GET  /reports/category    → [{category, total}]
GET  /reports/export.xlsx → Excel download

Tests: one pytest test per endpoint

README — what to highlight
1. Screenshot description of the dashboard
2. 'Import 1 file, see errors by row number'
3. Why pandas for reports, SQL for row queries
4. How to run: docker compose up / uvicorn
5. Tests and how to run them
6. What I would add next (auth, caching)