Maplebear Inc.

Intuit, Inc.

Ji Chen

Houtao Deng

Zi Wang

Yijia Chen

Soren Zeliger

Ganesh Krishnan

Jagannath Putrevu

Deepak Tirumalasetty

Mingzhe Zhuang

Site Wang

Wa Yuan

Christopher Shey-Tau Sun

Michael Scheibe

Yile Liu

Sambarta Dasgupta

Aman Jain

Debasish Das

Sricharan Kumar

Trace Levinson

Zhaoyu Kou

An online concierge system maintains a data store with discrete time intervals for fulfilling orders and information on shoppers categorized in a tree structure based on their characteristics. When an order is received from a user's client device, the system employs a machine learning model trained on historical data to estimate the number of available shoppers and their corresponding levels in the tree structure. The order is tagged based on the items it contains, and these tags are matched to the levels of capable shoppers. Shoppers are grouped and scored based on their levels, and a group with the minimum score and capability to fulfill the order is selected. A shopper from this group is then dispatched for order fulfillment, leading to a decrement in the estimated number of shoppers in the selected group. This approach optimizes the allocation of shoppers to orders, ensuring efficient and timely fulfillment.

Allocating shoppers for order fulfillment by an online concierge system accounting for variable numbers of shoppers across different time windows and varying capabilities for fulfilling orders

An online concierge system assigns shoppers to fulfill orders from users. To allocate shoppers, the online concierge system predicts future supply and demand for the shoppers' services for different time windows. To forecast a supply of shoppers, the online concierge system trains a machine learning model that estimates future supply based on access to a shopper mobile application through which the shoppers obtain new assignments by shoppers. The online concierge system also forecasts future orders. The online concierge system estimates a supply gap in a future time period by selecting a target time to accept for shoppers to accept orders and determining a corresponding ratio of number of shoppers and number of orders. The online concierge system may adjust a number of shoppers allocated to the future time period to achieve the determined ratio number of shoppers and number of orders.

Predicting shopper supply during a time interval based on interactions by shoppers with a shopper assignment application during earlier time intervals

An online concierge system receives two types of orders, one of which requires fulfillment in a specific time interval, while the other can be fulfilled anytime up to a specific time interval. A machine learning model, trained on historical data about available shoppers in discrete time intervals, is used to predict how many shoppers will be available to fulfill orders in each time interval. For each time interval, the system retrieves the relevant orders of both types and creates candidate groups including orders of both types. For each group, the system determines a fulfillment cost based on items in the orders. The candidate group with the lowest cost is selected, and the orders in the selected group are sent to devices of available shoppers in that interval, prompting the shoppers to view and fulfill the orders.

Allocating shoppers and orders for fulfillment by an online concierge system to account for variable numbers of shoppers across different time windows

An online concierge system allows users to order items within discrete time intervals later than a time when an order was received or for short-term fulfillment when the order was received. To account for a number of shoppers available to fulfill orders during different discrete time intervals and numbers of orders for fulfillment during different discrete time intervals, the online concierge system specifies a target rate for orders fulfilled later than a specified discrete time interval and a threshold from the target rate. A trained machine learning model periodically predicts a percentage of orders being fulfilled late, with an order associated with a predicted percentage when the order was received. The online concierge system increases a price of orders associated with predicted percentages greater than the threshold from the target rate. The increased price of an order is determined from a price elasticity curve and the predicted percentage.

Adjusting demand for order fulfillment during various time intervals for order fulfillment by an online concierge system

Certain aspects of the present disclosure provide techniques for performing finite rank deep kernel learning. In one example, a method for performing finite rank deep kernel learning includes receiving a training dataset; forming a set of embeddings by subjecting the training dataset to a deep neural network; forming, from the set of embeddings, a plurality of dot kernels; linearly combining the plurality of dot kernels to form a composite kernel for a Gaussian process; receiving live data from an application; and predicting a plurality of values and a plurality of uncertainties associated with the plurality of values simultaneously using the composite kernel.

Finite rank deep kernel learning with linear computational complexity

An online concierge system allows users to order items within discrete time intervals later than a time when an order was received. Each order may require a different set of characteristics for fulfilment by shoppers. Because different shoppers may have different capabilities, it is most efficient to reserve shoppers with specialized characteristics for orders that require them. The online concierge system maintains a set of hierarchical structures for different characteristics of shoppers, with each level in a hierarchical structure having a value. To select a shopper to fulfill an order, the online concierge system scores identifies groups of shoppers having characteristics capable of fulfilling the order based on levels in the hierarchical structure for each characteristic of a group. A shopper from a group having a minimum score is selected to fulfill the order.

An online concierge system allows users to order items within discrete time intervals later than a time when an order was received. The online concierge system allocates a specified percentage of an estimated number of shoppers during a discrete time interval to fulfilling orders received before the discrete time interval. An order may include a flag authorizing flexible fulfillment of the order along with a discrete time interval, which allows the order to be fulfilled earlier than the identified discrete time interval. The online concierge system generates groups of multiple orders authorizing flexible fulfillment and determines a cost for fulfilling different groups of orders. The online concierge system identifies a group of orders authorizing flexible fulfillment having a minimum cost for fulfillment by a shopper, allowing for more allocation of shoppers by enabling grouping of orders identifying different discrete time intervals.

An online concierge system displays an interface to a user identifying an estimated time of arrival for an order. To generate the estimated time of arrival for the order, the online concierge system trains a prediction engine to predict delivery time based on a predicted selection time for a shopper to select the order for fulfillment and predicted travel time for the shopper to deliver items of the order to a location identified by the order. The online concierge system generates a policy optimization model that computes an adjustment for the predicted delivery time. The adjustment is determined by solving a stochastic optimization problem with a constraint on a probability of the order being delivered after the estimated time of arrival. The predicted delivery time combined with the adjustment determines the estimated time of delivery displayed to the user to balance between minimizing late deliveries and wait times.

Determining estimated delivery time of items obtained from a warehouse for users of an online concierge system to reduce probabilities of delivery after the estimated delivery time

Growing community of inventors

Mountain View, CA, United States of America

Ji Chen