Interview question

You have bookings(booking_id, route, booked_at) and a table of fare snapshots fare_snapshot(route, snap_ts, fare) that is appended whenever the price engine publishes a new fare, a few times per day per route. For a training table you need the fare that was visible at the moment of each booking. Write the query (or the pandas call), and say what can go wrong in production.