Shaping tomorrow: From P2P displays to city simulations
Advanced technology is vital to modern life, from road safety alerts and driver assistance to simulations of stores, financial systems and even entire cities. These innovations make our lives safer and more comfortable, yet we rarely notice or think about the engineers and programmers behind them. Case in point: The solutions created by DXC Luxoft's Automotive Clusters team.
Who creates the innovations that the whole world uses? We asked Arthur Shaikhatarov, AI/ML project manager and DXC Luxoft’s AI subject matter expert, to talk more about this cutting-edge team.
Team History
DXC Luxoft's Automotive Clusters have a global presence, with more than 200 professionals working across Serbia, Poland, India, Brazil, the U.S., and Mexico. It all started in 2019 when four teams joined forces: Artificial Intelligence/Machine Learning (AI/ML), Embedded Systems, Graphics, and Simulation.
We began with some R&D, studying business problems and identifying technologies likely to be profitable. Soon afterward, we created Smart Room, an application that lets users model changes to interior spaces in a single click. You just take a photo, and you’ve already painted the walls, replaced the sofa, and added a carpet. There are many similar solutions these days, but back then we were pioneers, using state-of-the-art algorithms.
Soon, we started receiving requests for techniques no other companies had used. That's when our team began tackling non-standard tasks beyond traditional software development.
Domains and technology stacks
Artificial Intelligence/Machine Learning
The AI/ML domain has its own subcategories. Below are a few of them, with examples of projects we’ve completed.
Computer Vision
We develop image and video analysis technologies, like our EfficientDet-based car DVR detector, which identifies road signs and potholes using both real and synthetic data. This is crucial for self-driving vehicles, as safety demands noticing every detail.
We also enhanced the FaceNet facial recognition system with Fast MTCNN, boosting accuracy to nearly 94% and cutting training time by 35%.
Data Science
Data scientists turn numbers into powerful tools for optimizing processes. Take our “store digital twin” solution for a retailer with a network of 5,000+ stores worldwide. Our task was to optimize processes so that customers would spend less time in line. We created a store simulation using ML, Unreal Engine, and C++ to test hypotheses, which resulted in shorter lines and increased profits.
Generative AI (GenAI)
We use GenAI to make life easier. For example, we employed the LLaVa model to find anomalies in car graphical user interfaces (GUIs). What once took testers hours is now automated, with AI spotting bugs like misaligned buttons or incorrect text.
Machine Learning Operations (MLOps)
Turning chaos into order! We customize AI experimentation infrastructure, so customers don't have to worry about crashes and bugs. Our Grafana and Kibana-based interactive dashboards monitor performance and troubleshoot problems. For Airflow and MLflow experiments, we convert AI models to different formats like ONNX, ensuring smooth, efficient operation on any system.
Natural Language Processing (NLP)
We develop technologies that allow computers to interpret human language. Imagine a system that reads the news, analyzes its sentiment and predicts how stock prices will change. You don’t have to imagine it — we built it using BERT and CatBoost.
Embedded Systems
In this domain, we apply low-level programming to create solutions for low-power devices such as cameras and sensors. One example is our pillar-to-pillar (P2P) display for cars. Powered by Android Automotive, it allows users to control vehicle functions, watch movies, video conferences, and even plan their next vacation. Learn more about this project on YouTube.
EdgeAI
The subcategory of EdgeAI involves AI working with embedded systems. Imagine a driver monitoring system that not only watches the road but monitors how you're feeling behind the wheel. If you start yawning or stay still for too long, the system reminds you to take a break. It's all powered by a tiny chip and camera mounted right in front of the driver.
Graphics
Our artists and graphic designers work with 2D and 3D objects using state-of-the-art tools like Unreal Engine, Blender, and Maya to bring our digital solutions to life.
We regularly help create GPU tech demos, provide test materials, and share our expertise. We also generate synthetic data to train AI filters like Super Resolution and Denoiser from the FSR suite, which are available in modern game engines (Unreal, Unity, etc.) and software for popular consoles.
In one project, we showcased VR streaming tech enabling users in VR headsets to control a remote, high-quality renderer with a mock car brand configurator.
Simulation
Here, virtual worlds are created that bridge the gap between the real and the digital. In our autopilot training project, we simulated a real city, with actual streets, buildings, traffic lights, and pedestrians. AI algorithms were trained to control a car without driving on real roads, saving resources and, more importantly, enhancing safety. A nice bonus: We didn’t have to deal with traffic jams or road rage.
Modern tools and technologies
Our collaboration with leading hardware manufacturers gives us access to the latest technologies. In Serbia, we have our own server room with a GPU cluster, letting us work on graphics, simulation, and AI/ML efficiently. Meanwhile, our exceptional data security and in-house computing power help us protect client information and maintain high standards.
A place where you want to grow
At DXC Luxoft, “career growth” isn’t just an HR buzzword. Every one of our employees has access to our Training Center's catalog, with courses ranging from linear algebra to game development. Our Automotive Clusters team even has its own internal training programs and we partner with universities where our employees can teach classes, find talented students, and train them for our projects.
At DXC Luxoft, you can even change career tracks. Some employees have transitioned from one role to another quite quickly — from interface developers to machine learning specialists in just one month — balancing reskilling with their main jobs thanks to our strong training infrastructure.
Our team spirit
Our team includes professionals with rare competencies, but it’s not just hard skills that make us so productive.
We foster a friendly atmosphere where we can learn and freely exchange ideas. Every Friday, our team gathers to discuss news and projects and socialize. We celebrate holidays together with quizzes and board games. (Our board game president knows every game ever created!) New project ideas often emerge from these meetings, and any team member can pitch them to management.
Join us to change the world!
With innovative projects, advanced technology, a welcoming culture, and endless ways to grow, working at DXC Luxoft is a great opportunity to learn new skills and implement your most ambitious ideas. Check out our career opportunities and become part of a company that’s shaping the future of technology!
