The modern world is changing at a pace that few generations before us have experienced. Technologies that were once confined to elite research laboratories and multinational corporations are today becoming accessible to local teams, startups, and young researchers. Artificial intelligence, data science, and high-performance computing (HPC) are no longer abstract buzzwords. They are practical tools shaping how products are built, how knowledge is created, and how careers are formed.
Access to high-performance computing (HPC) infrastructure has a direct impact on the competitiveness of small and medium-sized enterprises (SMEs) nowadays.
Here is why this matters in practice.
- Speed becomes strategy: HPC reduces simulation and processing time from days to hours. SMEs can prototype faster, test more scenarios, and shorten time-to-market. In sectors such as energy modelling, biomedical engineering, or advanced manufacturing, this difference is decisive.
- Complexity becomes affordable: Without HPC, advanced modelling, AI training, and large-scale data analytics remain out of reach.
- Innovation becomes scalable: HPC supports AI development, big data processing, and engineering simulations at scale. SMEs can move from proof-of-concept to industrial application without outsourcing core knowledge.
Companies that rely on validated simulations and data-driven design gain stronger positioning in EU projects, tenders, and international partnerships. Access to HPC infrastructure strengthens technical maturity. In the EU context, initiatives such as EuroHPC Joint Undertaking and national competence centres expand access for SMEs. In Bosnia and Herzegovina, participation in the National Competence Centre for High-Performance Computing in Bosnia and Herzegovina ecosystem enables local companies to experiment, train staff, and pilot advanced computational solutions.
ARTI Analytics: Building an agent-orchestrated, no-code AI Platform for complex workflows
Such a platform goes far beyond static machine learning pipelines. It relies on multiple multimodal intelligent agents that interact, evaluate results, adapt strategies, and continuously optimize outcomes. This approach dramatically increases development complexity and, consequently, demands substantial computational resources to support experimentation, large-scale model training, and rapid validation cycles.
The DSML (Data Science and Machine Learning) platform was developed with a clear guiding principle: to enable teams to design, train, and deploy AI models more efficiently, reliably, and transparently, even when dealing with complex datasets and demanding use cases. A particular focus of the platform is its computer vision module, a cornerstone technology in many modern applications, including healthcare, industrial inspection, smart cities and security.
Developing computer vision solutions, however, is far from trivial. Training such models typically involves large image datasets, repeated experimentation with different model architectures, extensive hyperparameter tuning, and continuous validation to ensure robustness and generalization. On conventional hardware, these processes are time-consuming and often force teams to make compromises, either in model quality or in development timelines. This is where high-performance computing fundamentally changes the equation.
From NCC HPC BiH Infrastructure Access to AI Platform Training and FORTISSIMO Call Application
ARTI Analytics approached the National Competence Centre for High-Performance Computing in Bosnia and Herzegovina with a clearly defined product vision and prior technical expertise. In line with the NCC HPC BiH methodology, the company was classified as a mature user — a team that understands HPC concepts, parallel workloads, and the computational implications of large-scale AI development.
An initial technical assessment was conducted to evaluate workload characteristics, scalability needs, and expected development trajectories. Based on this assessment, ARTI Analytics was granted structured access to the experimental HPC infrastructure located within Verlab Institute and allocated dedicated compute time tailored to their experimentation and training requirements. The team transitioned from local resource constraints to an environment capable of supporting high-parallel model training, controlled benchmarking, and uninterrupted execution of demanding AI workloads. Instead of sequential experimentation limited by local hardware, ARTI Analytics leveraged large-scale parallelism to train and evaluate multiple model configurations simultaneously under reproducible conditions.
Key outcomes included:
- Significant reduction in training and experimentation time
- Acceleration of iteration cycles for model pipelines and preprocessing strategies
- Reliable benchmarking across identical computational environments
- Improved model stability and reproducibility for enterprise deployment
Time-to-market for the DSML computer vision module was substantially shortened. Validation and optimization cycles were compressed without compromising robustness. The expanded experimentation space strengthened model performance and reliability.
From proof of concept to strategic positioning
Following the successful Proof of Concept developed on the HPC infrastructure, NCC HPC BiH worked in parallel with ARTI Analytics to identify suitable European funding and acceleration opportunities. Within this process, the team jointly identified the FORTISSIMO call as a strategic pathway for scaling and industrial validation. Preparation of the application was conducted in close collaboration, aligning technical maturity achieved through HPC access with the programme’s objectives of supporting advanced digital innovation in SMEs.
HPC access therefore evolved into something larger:
- Technical validation
- Market acceleration
- European positioning
The trajectory moved from infrastructure access, to PoC validation, to international programme submission.
Opportunities for Young Generations
Perhaps the most important long-term impact of this collaboration lies in its relevance for young professionals, students, and early-career researchers. Today’s labor market increasingly demands not only theoretical knowledge, but also practical experience with advanced tools, scalable systems, and real-world constraints.
Through initiatives like training on HPC infrastructure, young participants gain the opportunity to:
- Work on enterprise grade products, not just academic assignments
- Develop hands-on experience with HPC systems and advanced compute heavy workflows
- Understand the entire lifecycle of AI development, from data preparation to deployment
- See how their skills translate into tangible value for industry and society
This is particularly significant in the context of Bosnia and Herzegovina and the wider Western Balkans. While many young talents seek opportunities abroad, access to modern research and development infrastructure at home demonstrates that high-level innovation is possible locally, given the right ecosystem and partnerships.
Collaboration as a Model for the Future
The partnership between the DSML team and NCC HPC BiH, led by the Verlab Institute, illustrates a model of collaboration that is increasingly essential in the digital era. By bringing together research infrastructure, private-sector development teams, and motivated young professionals, such initiatives create a fertile ground for sustainable innovation. High-performance computing, in this context, is not an end in itself. It is an enabling platform that allows ideas to mature faster, reduces uncertainty in development, and supports informed decision-making throughout the innovation process.
In a world increasingly shaped by data, algorithms, and computational power, initiatives like this are no longer optional. They are essential. The collaboration with NCC HPC BiH and the Verlab Institute stands not only as a successful project, but as a blueprint for how emerging technologies can shape the modern world and open new horizons for future generations of innovators.
Written by Tarik Hubana, PhD, ArtiAnalytics and Adna Softić, MSC, Verlab Institute
