Price suggestions for secondhand listings

Design the price suggestion shown when someone lists an item on a secondhand marketplace for furniture and electronics. The seller enters a category, brand, condition and age, and should see a suggested price range within half a second. The marketplace gets about 50,000 new listings a day and has 2 million sold listings from the past two years with their final sale prices. The data science team will build a regression model; you design how it is trained, served and monitored. Prices drift with the seasons and with new product releases, and many sellers ignore the suggestion.