As Lexmark moved beyond printing and imaging, the company faced the many challenges of digital transformation. In particular, Lexmark understood that vast amounts of siloed data existed across business applications, making actionable insights from the data extremely difficult to uncover.
The company invested heavily in the Internet of Things (IoT) to drive business outcomes, resulting in significant results for its customers and itself. Lexmark's success in leveraging IoT stood out from the pack, in comparison to 84% of companies stuck in “pilot mode” working on IoT projects (McKinsey). With this finding, Lexmark set out to build a cutting-edge solution that could help other companies open the floodgates to their own digital transformations.
Lexmark partnered with Differential to bring this vision to life, exploring edge computing to digitally transform the way businesses operate and make data-informed decisions.
We designed the platform to offer powerful analytic capabilities, including optimized big-data time-series processing and a simplified UI for building queries and exploring data. The analytics engine supports SQL syntax and enables users to gain insights through intuitive dashboards.
The technical infrastructure is designed for hyper-scalability, leveraging AWS Lambda for serverless computing, FaunaDB for scalable data storage, and AWS Timestream for capturing time-series data. The platform also uses Docker-in-Docker (DinD) to share properly versioned and secured device skills as docker images, running on AWS Fargate on a dedicated EC2 Cluster. We used a multi-cloud architecture, with some components running in Azure, including Azure IoT Hub and Azure Device Provisioning Service (DPS).
Providing real-time telemetry data was important to enable users to monitor the health of their devices and quickly detect and address any issues that may arise. This telemetry data is sent to the cloud for processing, allowing users to see when devices are offline or when failures occur. Additionally, users can directly access their devices via a reverse proxy connection and run ad-hoc scripts from any terminal.
To simplify infrastructure management, we worked with Pulumi (our premium choice for Infrastructure-as-Code). This enables consistent configuration across multiple infrastructure providers and environments, and makes deployments a simple one-command/one-click experience. With these capabilities, the edge computing platform can manage a range of use cases.
Not only is the team highly skilled and capable of turning out novel work at a reasonable cost, but they have a work process that brings a startup mentality to enterprise-level digital products. Their approach is lean, effective, and gets working prototypes in their customer's hands quickly.
The Final Product
Meet Optra. A first-of-its-kind integrated hardware and software platform for rapidly deploying edge AI applications securely and at scale.
Optra offers a range of features for managing a large fleet of connected devices. One of its primary capabilities is the ability to create and deploy machine learning (ML) models using raw data collected from these devices. The platform analyzes, moves, and/or exposes data produced from those devices. With the push of a button, users can easily deploy these models across their entire fleet of devices, regardless of their location.
With the edge computing capabilities of Optra, businesses save time and money with fewer servers and less infrastructure.
While Optra has been designed to scale for any business case, the platform has skills and product offerings geared towards specific industries.
Automate human visual inspection processes. Optra’s comprehensive I/O capabilities makes it an ideal device for controlling external hardware.
Operations visibility and people & process traceability. Configure thousands of devices across a chain of retail locations - all from a single portal. makes it an ideal device for controlling external hardware.
Vehicle traffic, identification, & incident awareness. Optra’s advanced GPU is ideal for vision-based Machine Learning tasks like counting cars and parking spaces.
Do you have an idea for your next digital product?
We’d love to connect about your project.