Step 1: Generate POS Transactions

First, we'll create a synthetic POS data generator that produces realistic retail transactions. In production, this would be replaced by your actual POS system feed.

The Generator

input:
  generate:
    interval: 100ms
    mapping: |
      # Unique transaction ID
      root.txn_id = uuid_v4()

      # Store and terminal
      root.store_id = random_int(min: 1, max: 50)
      root.terminal_id = random_int(min: 1, max: 10)
      root.timestamp = now()

      # Transaction type (weighted: 85% sale, 10% return, 5% exchange)
      let roll = random_int(min: 1, max: 100)
      root.type = match {
        $roll <= 85 => "sale",
        $roll <= 95 => "return",
        _ => "exchange"
      }

      # Payment method
      root.payment_method = ["card", "card", "card", "cash", "mobile", "gift_card"]
        .index(random_int(min: 0, max: 5))

      # Employee/cashier
      root.employee_id = "EMP-" + random_int(min: 1000, max: 9999).string()

      # Basket items (1-6 items per transaction)
      let categories = ["grocery", "produce", "dairy", "bakery", "meat",
                         "frozen", "beverage", "household", "personal_care", "electronics"]
      let item_count = random_int(min: 1, max: 6)
      root.items = range(0, $item_count).map_each(
        {
          "sku": "SKU-" + random_int(min: 10000, max: 99999).string(),
          "name": "Item " + random_int(min: 1, max: 500).string(),
          "category": $categories.index(random_int(min: 0, max: 9)),
          "qty": random_int(min: 1, max: 4),
          "unit_price": (random_int(min: 99, max: 4999).number() / 100)
        }
      )

      # Calculate totals
      root.subtotal = root.items.map_each(this.qty * this.unit_price).sum().round(2)
      root.tax_rate = 0.0875
      root.tax_amount = (root.subtotal * root.tax_rate).round(2)
      root.total_amount = (root.subtotal + root.tax_amount).round(2)

      # Returns are negative
      root = if root.type == "return" {
        root.assign({
          "subtotal": -root.subtotal,
          "tax_amount": -root.tax_amount,
          "total_amount": -root.total_amount
        })
      }

Test It

Save the config and pipe output to stdout:

output:
  stdout:
    codec: lines

expanso-edge run --config pos-generator.yaml

You should see transactions flowing:

{
  "txn_id": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
  "store_id": 42,
  "terminal_id": 7,
  "timestamp": "2026-02-11T04:00:00.000Z",
  "type": "sale",
  "payment_method": "card",
  "employee_id": "EMP-4521",
  "items": [
    {"sku": "SKU-34521", "name": "Item 127", "category": "grocery", "qty": 2, "unit_price": 3.49},
    {"sku": "SKU-78234", "name": "Item 303", "category": "produce", "qty": 1, "unit_price": 5.99}
  ],
  "subtotal": 12.97,
  "tax_rate": 0.0875,
  "tax_amount": 1.13,
  "total_amount": 14.10
}

What's Happening

At 10 transactions per second across 50 stores, this simulates ~43 million transactions per day — a realistic volume for a mid-size retail chain. Each transaction includes:

• Basket-level detail: Individual SKUs with categories, quantities, and prices
• Weighted distribution: 85% sales, 10% returns, 5% exchanges
• Calculated fields: Subtotal, tax, and total with proper rounding
• Negative amounts for returns: Returns flip the sign on all monetary fields

Next Step

The raw data is flowing, but it's missing store context. Next, we'll enrich each transaction with store metadata.

→ Step 2: Enrich with Store Metadata