Smed.x vision is create innovative solutions in medicine, finance and industrial research taking advantage of the advances in machine learning algorithms and high performance computing.
Whether it is our 3D reconstruction and image processing algorithms that power augmented reality medical products, or our machine learning expertise helping money management companies to do analysis faster and more efficient than ever before, smed.x team of scientists, developers and product managers will make your product vision a reality, however impossible it may seem to others.
Constantine Bychenkov, CEO
Our aim — bringing meaningful innovations to our customers in global markets. Please contact me to discuss how our talented team can support you to do your business to the best of your abilities.
In close cooperation with clients, smed.x develops custom technologies and software solutions in the fields of image processing, 3D visualization and scene reconstruction. Using the smed.x core technologies we can provide state of the art base technologies that can enhance and be integrated into many current frameworks. Some of the technologies we are proficient with in addition to our in-house core technologies include ITK, TubeTK, VMTK and general-purpose CUDA and OpenCV.
For medical applications, smed.x created best-in-class solutions for enhanced 3D reconstruction of specific human organs (CT, MRT, X-Rays and ultrasound imaging). Our experience also includes 3D reconstruction from multiple images.
GPU acceleration leads to a tremendous performance boost for many applications in finance, such as portfolio risk analysis, derivatives pricing, portfolio index-tracking error/cost optimization, Monte-Carlo simulations, back-testing of trading strategies, etc. On the other hand, researchers and software engineers with solid algorithmic background, in-depth knowledge of common tools (R, MatLab, SAS, SPSS, NumPy/SciPy), GPU programming experience and financial industry expertise are hard to find.
Taking advantage of the GPU-accelerated libraries, such as FFT, BLAS, sparse matrix operations, RNG, among others deliver significant speedups when compared to CPU-only libraries. When you have an algorithm that is not available among these libraries, smed.x experts can vectorize and optimize it for GPU architecture.
HPC and GPU software engineering is the practice of optimizing software to enable it to perform better on modern multicore processors, GPUs, HPC clusters and supercomputers. This may include optimizing the algorithm to make it more scalable, improving the code to increase hardware utilization to improve performance, or port the code from CPU-based HPC to GPU-based clusters.
smed.x parallel programming experts will work with you to port your proprietary software to CPU-GPU clusters, by introducing parallel processing into your applications using CUDA, OpenMP, MPI and other appropriate high performance technologies.
smed.x team can help you design your own high performance cluster, deploy it on your premises or at one of our high-density colocation partners, and optimize your algorithms or help you develop a new ones to get the best utilization of resources, whether it is GPU accelerator cards, CPU cores, memory bandwidth or inter-processor communications. We can deliver the software that can scale from a small consumer-grade GPU card to a production cluster with hundreds of servers in a cluster built with the latest technology. With smed.x, you can be sure that your domain expertise is supported by the most capable technical computing available on the market today.
Embedded High Performance Computing is the next frontier where GPUs can help accelerate the pace of innovation and deliver significant benefits in the fields of computer vision, robotics, automotive, image signal processing, network security, medicine, and many others.
In-house hardware lab that includes Jetson TK1 Development Kit allows smed.x experts to rapidly develop GPU-accelerated embedded applications, bringing significant parallel processing performance and exceptional power efficiency to embedded applications.