More Than Just Hype: Jakub Dunak on Building AI Systems People Can Trust

AI Systems Development: Ensuring Reliability and Compliance

Artificial intelligence is transitioning from experimental phases to practical application. However, challenges such as weak infrastructure, poor data quality, and compliance issues hinder its progress. Jakub Dunak, Principal Architect at Ness Digital Engineering, emphasizes that trust is as vital as speed in AI development.

In the second quarter of 2025, global expenditure on cloud infrastructure reached $95.3 billion, according to Canalys. A significant portion of this is attributed to businesses expanding AI capabilities. Industries such as healthcare, banking, and e-commerce are leveraging AI for diagnostics, fraud detection, and enhancing online experiences. Despite the growing demand, challenges in ensuring reliability, safety, and compliance of AI models persist.

Jakub Dunak's Experience in AI Systems

Jakub Dunak's extensive experience includes safeguarding Europe's financial systems at Diebold Nixdorf and developing multi-cloud strategies at CANCOM Slovakia. He has also led AI infrastructure projects at Ness Digital Engineering. His credentials include AWS Solutions Architect Professional and Microsoft Cybersecurity Architect Expert certifications.

Ensuring System Reliability at Diebold Nixdorf

At Diebold Nixdorf, a primary focus was on system reliability. Managing databases for major banks and retailers required minimizing downtime to prevent disruptions in ATMs and sales systems. Disaster recovery protocols and automated backups were implemented to enhance system resilience. These measures reduced recovery times significantly and reinforced the importance of building robust systems.

Managing Multi-Cloud Complexity at CANCOM Slovakia

At CANCOM Slovakia, the challenge was managing the complexity of multi-cloud environments. Different cloud platforms presented unique challenges in security, compliance, and billing. To address this, Jakub developed security controls compatible across platforms, aligned compliance rules, and optimized workload distribution. This approach minimized downtime risk and enhanced system manageability.

Compliance and Speed at Ness Digital Engineering

At Ness Digital Engineering, compliance is integrated into the design of AI systems from the outset. By incorporating encryption, access rules, and data residency into core designs and leveraging automation, the company ensures swift and compliant scaling. This strategy reduces infrastructure costs and simplifies audits.

Ensuring data quality is critical for AI reliability. Techniques such as data lineage frameworks, machine learning tools for detecting incomplete data, and bias mitigation strategies are employed to maintain data integrity. An example includes expanding datasets and conducting fairness checks to improve model reliability.

To address energy consumption, AI infrastructure utilizes scalable systems like Kubernetes to adjust capacity based on demand. Workloads are distributed to regions with efficient energy usage, reducing carbon footprint and operational costs.

By 2030, companies will need to adopt zero-trust security, compliance-as-code, and efficient workload management as standard practices. These measures will be essential for scalable and responsible AI deployment.

Based on reporting by TechBullion.