Design Amazon Locker: a package pickup locker network

Design a self-service package locker network. At checkout a customer picks a locker location; when the package arrives, the system assigns a compartment big enough for it and sends the customer a code. The customer has a few days to collect it, after which it is returned.

Requirements

• Functional: show nearby locker locations with free capacity at checkout; reserve capacity for an order; assign a compartment of the right size when the driver arrives; generate a one-time pickup code; open the compartment for a valid code; handle expired pickups and returns; support drivers loading many packages at once.
• Non-functional: a compartment is never given to two packages; lockers keep working when their network connection drops; codes are hard to guess; status is accurate enough for checkout to rely on.

Clarifying questions

• How many locations and compartments, in how many sizes?
• When is a compartment reserved: at checkout or at delivery?
• How long does a customer have to collect, and what happens after?
• Can customers return items through the locker too?
• What hardware and connectivity does a locker have?

What a strong answer covers

• A data model: locations, compartments with size and state, reservations, deliveries, pickup codes.
• Capacity planning at checkout by size class per location and day, separate from assigning a physical compartment at delivery.
• Compartment assignment: smallest compartment that fits, with an atomic state change so two deliveries cannot take the same one.
• The locker as an edge device: a local state copy and code list so it works offline, syncing events to the backend afterwards and resolving conflicts.
• Code generation and verification, rate limiting of attempts, and audit of door openings.
• Expiry jobs that schedule returns, notifications to customers, and monitoring of hardware faults.

Follow-ups

• A package does not fit the compartment assigned to it.
• A locker is offline for a day while customers keep collecting.
• A location fills up during the holidays. How do you avoid sending customers to a full locker?
• Add returns: a customer drops off an item for a refund.