Projects

Machine Learning is a broad area of computer science topics. Essentially the topic can be broken down into the two categories of model training and model implementations. Companies like Apple, Google and IBM are doing their best to make the technology more accessible for developers to leverage and add ML concepts into applications. The Core ML framework, along with the other related frameworks (like Vision), provided by Apple helps to simplify the model implementation side of ML. Core ML and Vision work closely together to process images and make predictions about image contents.

Computer vision is... pretty much exactly what it sounds like. It's the science of giving your hardware the ability to 'see' via a camera input. The implementation of machine vision is still under heavy research and we're passionate about being part of the advancement of the field. As this technology grows, the computational power required to execute the software grows with it. With Google and Apple introducing machine learning frameworks designed for mobile devices (Tensorflow Lite and Core ML), we're taking the initiative to incorporate them in our research and development.

Quickly calculating reports across large datasets can be a challenge in a traditional RDBMS environment. By leveraging an "inside out" data store like Kafka, we can durably store events, as well as transform them at scale and produce easily-queried reports on that data. We can even push updated data to clients by bridging Kafka topics to WebSockets or similar push notification mechanisms.

The temperature fluctuations within an office building are a constant source of employee complaints across all industries. Temperature monitoring IoT units allow for an administrator to keep track of these changes easily, but these solutions typically require a large amount of infrastructure to be built. Amazon's AWS IoT platform allows for the majority of the backend to be configured in a minimal amount of time and for a trivial cost.

This Innovation Lab project was done in collaboration with our client, Banners.com. They're a large-format printing company based out of Kensington, Minnesota. AndPlus created the software that merges a large number of graphic files into a single image that is print-ready. Occasionally, there are errors in the merge process that cannot be predicted manually. These errors are costly in that they hold-up the remainder of the process, causing a backlog of files that add up until the error is fixed. 

AndPlus was looking for ways to add some customer interactivity to their website. Utilizing a chatbot was the immediate choice for demonstrating knowledge within an emerging area and the ability to provide the appropriate user experience for customers. The sample chatbot was geared towards scheduling an appointment with an AndPlus employee. A web-based chatbot was preferred for testing, however Amazon Lex provides the ability to integrate with other chat-focused platforms like Facebook Messenger and Slack.

Image credit to AB Electrical & Communications Ltd. Manually managing deployments to staging and development environments eats up developer and operations time, especially when it is repeated multiple times during a sprint. Automating as much of the operational workload as possible allows developers to iterate faster and get better feedback from stakeholders by demonstrating in-progress work quicker, without having to delay user acceptance testing until a batch of work is complete. By using Kubernetes to automate deployments of new features, we can expose new demos faster than ever and decide if a feature is coming together the way we expect.

Augmented Reality (AR) is an area of computer science that allows programs to supplement the real world with artificial constructs. These constructs can take the form of static images, directions around an area, or even super imposing other environments on the existing one. By using AR, an app can simulate interactions with the real world in a virtual space. Android's solution to AR is ARCore, which lets most of the existing Android phones available execute AR apps with existing hardware. While this doesn't provide as much accuracy as custom hardware might provide (using the single camera on a phone compared to using two cameras to build a virtual understanding of the world), it makes AR technologies more accessible.

eCommerce reviews are crucial in spawning interest in products as well as driving sales. In the Amazon ecosystem users are able to post reviews on items that they may have never actually purchased. Several groups create fake reviews to drive down sales of technologies that they dislike (for example Sony fans creating fake reviews for Microsoft products). Amazon has tried several different techniques to disrupt this process, such as marking verified purchases, but the problem still continues.

Nowadays, almost every system you interact with is a distributed system. Rather than single machines running all of the code for an app or website, we have web apps, mobile apps, and API servers all interacting over the Internet. Adding new features to these apps often requires API changes, but introduces the risk of breaking older versions still being used. Our team introduced Pact as a way to enable teams to be more confident when building these types of systems.