A non-transitory computer-readable storage medium storing instructions for sorting a list of items to be picked in a warehouse, the instructions when executed causing a processor to:
- receive, from a first mobile application running on a computing device associated with a customer, data identifying a plurality of items to be picked by a picker in a warehouse;
- retrieve, for each of the plurality of items to be picked, an embedding describing the item, wherein an embedding comprises a vector that describes the item in a latent space;
- retrieve a machine-learned model that predicts a next item of a picking sequence of items, the model trained based on a plurality of sets of data, each set of data comprising a list of candidate items and a picking sequence in which the picked items in the list were picked, and wherein training the machine-learned model based on a set of data of the plurality of sets of data comprises:
- transmitting the list of candidate items of the set of data to a picker device associated with a picker of the plurality of pickers;
- receiving, from the picker device, the picking sequence for the list of candidate items;
- and
- retraining the machine-learning model based on the picking sequence by, for each item in the picking sequence:
- inputting, to the machine-learning model, an embedding for the item and a list of previously-picked items in the picking sequence to generate a prediction score for each unpicked item in the picking sequence, wherein a prediction score for each unpicked item represents a likelihood that the corresponding item is picked next by the picker;
- comparing the prediction score for each unpicked item to the item in the picking sequence;
- and
- updating a plurality of weights for the machine-learning model based on the comparison;
- receive, from a second mobile application running on a computing device associated with the picker, data identifying a first item of the plurality of items picked by the picker and a plurality of remaining items of the plurality of items;
- predict, using the retrieved model, a next item of the plurality of items to be picked by:
- calculating, for each of the plurality of remaining items, a score for the remaining item by:
- inputting the embedding for the remaining item into the machine-learned model;
- and
- receiving, from the machine-learned model, a score for the remaining item;
- and
- selecting the next item from the plurality of remaining items based on the calculated scores;
- and
- transmit, to the mobile application running on a computing device associated with the picker, data identifying the next item to be picked.