Nanografi Nano Technology


Materials Modelling for Nanocomposite Optimization

Numerous experiments are conducted for the optimized production of polymer-graphene composites by the company. This case study aims to reduce the cost and duration of the experiments through the material simulations using the National e-Infrastructure TRUBA.






Text Processing of Social Media Messages

Somera’s in-house developed systems process millions of social media posts and web pages for its clients and business partners. Their solutions require big data analysis and live streaming of the analysis results through dashboards. For the complete analytics workflow, Somera uses in-house servers as well as cloud resources. Within the scope of this case study, Somera will utilize National e-Infrastructure TRUBA to train and deploy AI-based models, porting components of its analytics workflow.




 Parabol Software


Public Transport Analysis Platform

Parabol has been carrying out R&D activities in the intelligent transportation systems sector. The company has also been developing a “Public Transport Analysis Platform” called Cermoni. This platform aims to analyze the passengers’ boarding data obtained through smart cards and the GPS location data of the vehicles, such as buses and trams, collected from the public transportation system of a city. Within the scope of this case study, Parabol will run some modules of the Cermoni on National e-infrastructure TRUBA’s HPC clusters. The company also aims to port an optimized, high-performance version of the Origin-Destination matrix computation module onto the HPC environment.





Image-Based Content Moderation Project

Machinetutors offers machine learning consultancy and custom AI services. This case study will try to solve real-time image-based content moderation using a custom image dataset crawled from the web with proper licenses and annotated by in-house tools. This training dataset comprises 600.000 images. Using the national HPC infrastructure TRUBA, the company will develop deep learning models to achieve highly accurate image categorization and minimize the adverse effects experienced by web users



Yapı Kredi Teknoloji


Fraud Detection with Graph Processing

Yapı Kredi Teknoloji aims to use machine learning (ML) to detect fraudulent bank transactions.  The goal is to provide the business unit with better predictions, reduce human effort, and increase the detection rate of accounts participating in fraudulent transactions. High-performance computing (HPC) capability is critical for effective experimentation and to build models based on graphs. NCC Turkey provides ML expertise and also resources from the national HPC infrastructure TRUBA for this case study.





Code Modernization for Glass Industry

Şişecam is one of the largest glass manufacturers in Europe. The company fabricates glass products using custom-made furnaces. To design these furnaces, they use a simulation framework that traces the glass particles. In a nutshell, the longer the particles stay inside the furnace, the better the quality of the product. For this purpose, they need to perform simulations with a large number of particles using their source codes; however, computational costs increase exponentially with an increasing number of particles.Within the context of this case study, NCC Turkey provides expertise in code optimization, parallelization, and modernization using national HPC infrastructure TRUBA







Simulation Optimization of a Patented Design with Parallel Computing on TRUBA Cluster

Design and Simulation Technologies (DSTECH) Inc. provides services for the solution of complex engineering problems encountered in different engineering disciplines such as environment, energy, and aerospace sciences. DSTECH intends to optimize a design patented by the European Patent Office for the efficiency enhancement of a disinfection system in potable water treatment plants. DSTECH develops an augmented simulation-optimization framework using Dakota, OpenFoam, and Python for the automatic optimization of the patented design.  With the assistance of NCC Turkey expertise, numerical simulations are performed on the national HPC infrastructure TRUBA.




Erste Software


HPC for Machine Learning and Digital Twin Health

Erste Software provides solutions for the software needs of the customers and manages many different R&D projects focused on IoT. In this case study, the data they need to deal with are streaming from around 10 production lines, each having multiple, 5-10 machines, where each machine is equipped with tens of sensors producing hundreds of data points in every second. They try to forecast future data points with a relative error threshold between prediction and observation and find anomalous points occurring rarely in the data set by using the HPC environment. Following, the company recommends the best maintenance points by using an artificial neural network on the national HPC infrastructure TRUBA.