Vision Computing platform at Target

Target’s purpose is to help all families discover the joy of everyday life. To bring this purpose to life, we work on core platform capabilities in vision computing and machine learning (ML). We enable unparalleled guest experience by deriving insights from structured data and unstructured content (videos, images and documents) using vision computing and machine learning platform capabilities. The platform provides an end-to-end, ubiquitous experience including acquiring, processing, analyzing and understanding content. It also helps in extraction of high-dimensional data in order to produce numerical or symbolic information, e.g., in the form of decisions. The machine learning platform enables ML developers to focus on writing efficient ML models and train/ deploy/ manage with minimal effort.

 These new capabilities were incubated in Target in India, Bangalore in 2020 right before the pandemic hit us. True to Target’s culture, this work is based on strong collaboration and partnership with globally distributed business and engineering teams. As much as the stories below are about the capabilities, the journey in the last 12+ months was equally about building teams. Nearly 60% of the team members supporting the build of these new capabilities joined Target in the middle of the lockdown, on-boarded virtually and continue to interact virtually.

 The below is a short narrative of all the products that are custom built using an open source stack guided by distributed micro-service architecture and cloud native.

Computer Vision services

We are building a highly scalable and reliable computer vision platform for Target. We want to leverage all vision IoT devices in the ecosystem (200,000 and growing) to derive insights and solve for inventory management, shortage, compliance, safety and many other retail operations. This can easily be extended to other retail domains like merchandizing, product design and management and supply chain.

Feature story – we leveraged AI and ML during the pandemic to count the number of guest (all customers in Target are called ‘guests’) using the camera at the entrance and exit. The solution was rolled out across more than 1500 locations using over 3000 cameras; the pilot rollout happened within weeks. This significantly helped meet the compliance needs of social distancing and guest metering at stores, ensuring the safety of our guests and team members. This saved thousands of man hours and possibly million of dollars.

Computer Vision hardware

This product is chartered to design custom cameras and build robotics solutions for specific vision-based business usecase. Currently, we are designing shelf camera to monitor shelves remotely and generate replenishment alerts when shelf space goes empty. This is an inter-disciplinary engineering discipline. The solution would require integrating with other Target platform like Vision, ML, IoT, Cloud environments and store facing apps. This will scale out to other IoT devices, robotics, AR (Augmented Reality) and wearables.

 Computer Vision data platform

This manages vision data from all distributed devices across thousands of sites. We expect to collect and process more than 50 million images this year across 200,000+ devices. The pipeline spans across data collection, management, curation/ annotation and democratization. We want to create largest ever Retail specific vision data.

 Machine Learning Platform

This focus on ML lifecycle management - from development to deployment at data center, on edge and at far edge. We enable ML engineers and data scientists to build and train models, conduct experimentation, tune for hyper-parameters, register in model registry, deploy and manage. The platform provides containerization support leveraging existing container architecture at Target. It also integrates with data engineering platforms for data management. Feature management capabilities are being integrated with vision embeddings.

It’s a long journey of a few years or more to build these capabilities and run at enterprise scale, considering the dynamically changing business landscape and advancements in AI as well as related technologies. Target continues to invest in building this capability and we are hiring !!

Join me in this journey. Apply to the positions below.

1. Director/ Senior Eng Mgr - Computer Vision Services
2. Director/ Senior Eng Mgr - Computer Vision Hardware
3. Principal Product Owner
4. PE (Principal Engineer) - ML Platform
5. Lead Engineer - ML platform
6. Sr Engineer - ML platform
7. Sr Engineer - Vision platform