NCC Admin, Author at HPC Montenegro

17.02.2026.17.02.2026.

PAID MNE Showcased in the EuroCC2/EuroCC4SEE Success Stories Booklet — Powering Smarter Trading with Supercomputing

We are proud to highlight PAID MNE as a featured success story in the EuroCC2 & EuroCC4SEE Booklet — demonstrating how HPC is transforming financial analytics and algorithmic trading.

At the heart of PAID MNE innovations lies PAID-T (Price Action Intelligent Detection Trading) — a smart trading platform that leverages advanced algorithms and AI/ML to dynamically adapt to market movements, optimise investment strategies, and manage risk with higher precision. Traditional computing systems quickly reached their limits. To unlock the required performance, the team scaled their solution to the LUMI supercomputer, one of Europe’s most powerful HPC infrastructures. By enabling multinode execution and real-time task distribution, PAID MNE achieved over 1.2 million simulations in under 5 hours — a process that previously would have taken days. This acceleration enables processing billions of historical transactions in hours instead of days, rapid identification of critical market patterns and data-driven optimisation/ increased accuracy of trading strategies.

This achievement, showcased through the EuroCC2/EuroCC4SEE project, demonstrates how supercomputing is becoming a powerful enabler of business innovation. PAID MNE’s journey is a clear example of how HPC and AI together can transform complex, critical data into faster, more profitable decisions.

16.02.2026.16.02.2026.

DeepMark Approved GPU Resources through AI Factory Playground Access

DeepMark is a deep-tech startup developing learning-based watermarking and provenance technology for AI-generated content. It previously used EuroHPC JU resources after receiving one year of access to the Leonardo Booster partition through a Development call. DeepMark leveraged these resources to advance next-generation watermarking research and benchmarking.

As the research deepened, the need for additional compute grew. Following consultations with the NCC Montenegro team at the University of Montenegro, DeepMark applied for GPU resources via the AI Factory Playground Access.

DeepMark has been awarded 5,000 GPU hours on the Leonardo Booster partition, granted for three months. This allocation will support work on robust, learning-based watermarking for AI-generated content—improving resilience to real-world edits and AI transformations and strengthening provenance and authenticity at scale. With this new allocation, DeepMark will expand its experiments and validation under real-world constraints.

11.02.2026.

Uhura Solutions Awarded Extension and Additional HPC Resources on Leonardo Supercomputer

The Uhura Solutions project “Generative AI Intelligent Process Automation Platform” has received a positive evaluation and has been granted an extension, including 4,500 node hours on the Leonardo BOOSTER HPC system for a 12-month period.

The additional HPC resources will support Uhura Solutions’ ongoing research and development in advanced Generative AI and Large Language Models (LLMs) tailored for the financial sector. Leveraging Leonardo’s state-of-the-art computing capabilities, the project will scale experiments from smaller open-source models to larger LLMs, enabling deeper research into optimization techniques such as parameter-efficient fine-tuning (PEFT), quantization, pruning, and model alignment.

The Generative AI Intelligent Process Automation Platform integrates fine-tuned LLMs, low-code development, and process automation workflows to significantly improve operational efficiency, reduce costs, and enhance customer experience in financial services. A key differentiator of the platform is its focus on industry-specific language and context, supported by carefully curated private datasets with strong attention to data quality, privacy, and compliance.

This extension and access to Leonardo BOOSTER represent a major milestone for the project. The additional computational power allows company to push the boundaries of financial-domain AI and accelerate the delivery of scalable, production-ready solutions for the European market

10.02.2026.10.02.2026.

Multi-GPU AI Train the Trainer Workshop

The Train the Trainer course was organized from 30 January to 5 February as a series of full-day, intensive lectures, delivered alongside the standard AI and multi-GPU computing courses. This workshop complemented the strong technical focus of the regular course with a dedicated educational track. While the regular course concentrated on hands-on skills for building, scaling, and optimizing AI workloads across multiple GPUs, the Train-the-Trainer track was specifically designed to equip future instructors with the knowledge and pedagogical tools needed to teach these topics effectively. With only a few HPC and AI centres across Europe offering such specialized education, this programme helps expand the network of trainers and institutions capable of delivering high-quality AI and multi-GPU courses at both national and European levels. Researchers from the National Competence Center (NCC) Montenegro participated in this course, further strengthening national capacities in advanced AI and multi-GPU training.

During the training event, participants developed a deep understanding of modern multi-GPU and distributed AI technologies, while also learning effective strategies for teaching these complex concepts. Guided by expert instructors, attendees explored state-of-the-art tools, frameworks, and best practices for scaling AI workloads, combined with pedagogical approaches, teaching materials, and hands-on experience tailored to support the delivery of their own courses. The lectures covered a broad spectrum of topics, including GPU architectures and access to HPC infrastructure; fundamentals of deep learning and the transition from CPU- to GPU-based workloads; distributed training with PyTorch (Distributed Data Parallel, model parallelism, PyTorch Lightning); large language models, fine-tuning techniques, and frameworks such as Hugging Face Accelerate and DeepSpeed; as well as computer vision, MLOps, Ray, Retrieval Augmented Generation (RAG), and hyperparameter tuning.

The training event was organised by NCC Poland, NCC Netherlands, NCC Hungary, NCC Belgium, NCC Italy, NCC Finland, NCC Sweden, all National Competence Centres for High-Performance Computing, CASTIEL2 and EuroCC2/EuroCC4SEE.

30.01.2026.31.01.2026.

NCCs from Italy, Denmark and Montenegro Share Experiences on Strengthening Industry Engagement in HPC, HPDA and AI

NCCs Montenegro, Italy, and Denmark held an online mentoring and twinning session (30.01.2026) focused on sharing valuable experiences and good practices in building initial industry contacts in the fields of HPC, HPDA, and AI. The discussion centred on practical approaches and lessons learned in engaging companies—particularly SMEs—and on transforming NCC technical expertise and resources into attractive, demand-driven service offerings.

The NCCs shared concrete examples of mapping local ecosystems and working with key national stakeholders, strategic partners and industry intermediaries—including chambers of commerce, industry clusters, digital agencies, science and technology parks, start-up incubators/ accelerators, etc., and European Digital Innovation Hubs (EDIHs)—to reach targeted companies and industry leaders. Collaboration models, including joint workshops, training activities, and project initiatives, were highlighted as effective means of promoting access to HPC and AI resources, expertise, and funding support.

The session also highlighted the importance of continuous visibility and coordinated outreach through channels such as LinkedIn, newsletters, and active participation in targeted conferences and networking events, as key enablers for promoting NCC services and strengthening industry uptake.

25.12.2025.25.12.2025.

Training event on Practical MPI Programming successfully completed

On 23 December 2025, at premises of the Faculty of Electrical Engineering, University of Montenegro, NCC Montenegro (EuroCC) successfully delivered a hands-on training session on MPI programming in Python using mpi4py. The event introduced participants to the fundamentals of the message passing paradigm for distributed-memory systems and explained how the MPI standard enables scalable parallel applications across multiple processes.

Hands-on training session on MPI programming in Python using mpi4py

The training combined concise theory with practical demonstrations. Core MPI concepts such as SPMD execution, communicators, process rank and size, and the basics of point-to-point communication using Send/Recv (including tags and common communication patterns) were covered. Participants also learned how to set up and run MPI programs with mpiexec, verify their environment, and interpret parallel output behaviour.

A key part of the session was a step-by-step example of parallelizing a numerical computation using the trapezoidal rule for integration. Through this case study, workload partitioning across processes and collection of partial results were demonstrated, along with a discussion of typical performance considerations such as synchronization overhead and potential bottlenecks at the root process.

23.12.2025.23.12.2025.

Train the Trainers Workshop @ NCC Montenegro: Agentic AI – From Core Concepts to Real-World Applications

Agentic AI is emerging as a powerful enabler of next-generation digital transformation, business automation, and real-time decision-making. Unlike traditional AI solutions, Agentic AI systems operate as autonomous, goal-driven agents capable of planning, reasoning, and adapting over time, making decisions without continuous human intervention, and collaborating with tools, systems, and dynamic environments. Recognising its growing relevance, NCC Montenegro organised a Train the Trainers workshop for NCC members, focused on the application of Agentic AI within experience-driven industries.

Train the Trainers Workshop @ NCC Montenegro: Agentic AI

The workshop was delivered by the NCC representative, Dr Armin Alibasic, who combined academic and industry expertise with his international and interdisciplinary experience across the automotive, airline, and theme park industries. The session highlighted the potential of real-time analytics enabled by Agentic AI, demonstrating how autonomous intelligence can support instant, data-driven decision-making. These concepts were illustrated through real-life industry applications and a hands-on demonstration on Databricks, a leading Data and AI platform.

The Train the Trainers approach addressed two complementary dimensions. The first focused on capacity building, providing participants with a structured and in-depth understanding of Agentic AI concepts, system architecture, required skills and tools, as well as the key challenges associated with designing and deploying agent-based AI solutions. These foundations were reinforced through business-oriented use cases, ensuring a clear link between theoretical principles and real-world industry practice. The second dimension emphasized knowledge transfer, equipping participants with the competencies needed to design, adapt, and deliver high-quality training activities tailored to SMEs, complex organizational systems, and strategic decision-makers.

By empowering trainers with both deep technical insight and practical training capabilities, NCC Montenegro ensures that Agentic AI knowledge can be effectively disseminated, scaled, and transformed into business value for industry, SMEs, and the broader digital innovation ecosystem.