Deep Fakes in real time
Deep Fakes are imitations of pictures and video, created using artificial intelligence. Common scenarios include exchanging faces in order to create the illusion to see another person in a video. Within the scope of a research project, the TNG Innovation Hacking team investigated Deep Fakes to better understand what is technically feasible. Furthermore, we wanted to explore the limitations of the technology - especially if and with what image quality it is possible to create Deep Fakes in real time.
Our research shows that it is possibled to exchange the face of a person filmed via a real-time video streamwith the face of another person, including applying facial expressions and movements of the person filmed.
By applying various techniques from the area of computer vision and neural networks, faces in the video feed are recognized, transformed and embedded in the video output. The project uses autoencoder networks trained in Keras. They were trained using so called GANs (Generative Adversarial Networks). Aditionally, the developers used different neuronal networks for face recognition and segmentation.
AI Size Recommender for Fashion
For a large fashion retailer, TNG developed an application that makes individual size recommendations for web shop customers. The task for the solution was – despite a large number of size ranges and partly low availability – to display to individual customers only those products really likely to fit. To do this, we started with an idea workshop, convinced our client by developing a prototype, and finally rolled out a production solution with measurable success. To create the individual size profiles, we experimented with various machine learning approaches. The final version was then implemented using AWS Lambda and Elasticsearch.
Launch of a Real Time Chat App
To strengthen the contact between customers and pharmacists we supported a wholesale pharmaceutical company in the development of a chat system.
As a "progressive web app", the chat is available without installation and is optimized for mobile devices. We employed cloud services and among many other services a GraphQL interface.
Users' feedback is collected and incorporated into the product, along with improvements and features, in an agile scrum process.
Modularization of a Monolithic Legacy Application through New Architecture and UI
We designed and verified a new architecture for a provider in the logistics sector, enabling incremental migration from two monolithic legacy applications to state-of-the-art modules. We supported risk reduction though various proofs of concept – particularly for the seamless integration of a modern React UI into the existing ExtJS application, as well as to verify backend integration mechanisms.
Introduction of a Cloud Microservice-Platform
We supported a client in building a microservice platform as part of a worldwide technology program. The platform is based on AWS and Kubernetes, which act as a blueprint for new service development and serve as a means to move the concern towards agile processes and Dev Ops.
The implementation of an automated, cloud-based CI/CD pipeline and templates for service development and deployment processes allow services to be delivered in a quality-controlled as well as cost-efficient way.
Payment process improvement
We helped a client meet the challenge of successfully processing all transactions even during periods with especially high order volumes, thus avoiding revenue loss.
To achieve this, we modularized existing code for connecting payment service providers and added two new providers.
The rejuvenated payment process now provides a failover for individual unavailable payment methods, thus delivering the required high availability while at the same time reducing costs.
TNG now Atlassian Platinum Solution Partner
In December 2018, having already been Atlassian Expert and Gold Solution Partner, TNG was accredited as an Atlassian Platinum Solution Partner, the highest level of Atlassian Partner. We have been using Atlassian products in our projects for 10 years as Atlassian Partners and can support our clients in all areas of digitalization with Jira, Confluence, Bitbucket etc. From requirements management through agile software development to optimization and automatic reports for Business Intelligence: our team of certified Atlassian experts can help you to reach your goals.
Customized Product Sorting
TNG developed a sorting algorithm for one of the largest European online shopping communities in the fashion market. The algorithm reorders the displayed products in real time, taking into account the community's buying behaviour as well as that of the individual customer. Visitors to the shop are shown the products that fit their individual size profile first, so that customers no longer need to use a size filter to find a suitable garment.
The size profile management can be accessed, thanks to AWS, via a simple REST API. To ensure the best possible performance, an elastic search plugin was also developed. This is installed directly in the shop system and is responsible for sorting the results list.
Development of a Trade Convention App
Using React and Node.js, we created a web application to accompany product sampling at trade conventions for a large engineering company.
Visitors were able to use the application on tablets or mobile devices. The display was optimized for various screen sizes using responsive design. Users rated various product qualities, while the currently selected rating was attractively visualized. Afterwards the application processed and analysed the answers, presenting users with a comparison of their rating with the average of all previous ratings.
Elastic Stack Upgrade
TNG performed an upgrade of the central Elastic Stack for company-wide logging at a large insurance company. This was a two-stage process, using an intermediate version to reach the latest version. It was the first upgrade of this system altogether and had to be executed manually, without central server management, on each server. At the same time, 6TB of production data in the stack was moved to new hardware. Despite these large scale changes, the effects on users were very small.