admin22 is a future-focused software engineering company that brings ambitious ideas to life. fintechview had an exclusive interview with Fedros Avraam, Chief Technology Officer at P.C ADMIN 22 LIMITED, discussing the agile approach the company follows, which ensures swift implementation of solutions, helping organisations stay ahead of the curve in a fast-paced digital world, especially now with AI.
AI will move from being a supporting tool to a core decision-making layer across industries.
Tell us about your company, the core services you offer, and the customers or market segments you serve.
I am part of a technology group focused on fintech and cybersecurity, operating across multiple regulated markets. Through our companies, we design, build, and operate secure, scalable financial platforms that enable businesses to accept payments, manage risk, and grow safely in an increasingly digital economy.
Our services span the full fintech lifecycle. On the financial side, we provide payment gateways, PSP and sub-acquiring services, virtual accounts, card payments, payouts and bulk payouts, QR payments, payment links, crypto payment processing, and e-wallet solutions. These services are offered through licensed entities in different jurisdictions, allowing us to support both local and international transactions.
Security is a core pillar of everything we do. Alongside fintech products, we offer cybersecurity services, including penetration testing, API and cloud security assessments, PCI DSS consulting, fraud prevention technologies, audit trails, and real-time transaction monitoring. This dual focus allows us to build systems that are not only innovative, but also resilient and compliant from day one.
Our target audience includes:
- Banks and financial institutions
- Payment service providers and fintech companies
- E-commerce platforms and marketplaces
- Crypto and digital asset businesses
- Enterprises handling high-volume or high-risk transactions
We work with clients that need more than a generic solution, businesses that require customization, regulatory awareness, strong security, and the ability to scale globally. Our goal is to act as a long-term technology partner, helping organizations deliver seamless financial experiences while staying secure and compliant.
Can you explain what an AI fraud detection tool is and how it works?
An AI fraud detection tool is a sophisticated risk assessment system that evaluates transactions in real time, estimating the likelihood of fraud rather than relying on static yes-or-no rules. Traditional systems block transactions based on fixed conditions, while our approach focuses on understanding patterns and probabilities. When a transaction is initiated, the system instantly analyses hundreds of behavioral and contextual signals, such as customer behaviour, transaction velocity, device consistency, and location patterns. These signals are processed by machine learning models that generate a fraud risk score on a scale from 0 to 100. Based on that score, merchants apply dynamic decision thresholds to approve, review, or reject transactions.
Finally, the system continuously improves through incremental learning. As new outcomes, such as confirmed fraud or chargebacks, are fed back, the model updates itself to reflect evolving fraud tactics. In short, it is not a static firewall; it is a dynamic system that becomes smarter over time.
Any challenges you faced during the development or implementation of an AI fraud detection tool?
As with any complex project, we faced a number of challenges along the way. One of the first was determining which model was best suited to the problem and how it should be configured. Equally important were decisions around data optimization: we engineered additional, high-value features derived from the existing data to enrich the model’s understanding. This process added domain knowledge directly into the learning pipeline, reinforcing the idea that a strong model begins and ends with well-structured, informative data. In addition, we incorporated user-driven signals to better capture and interpret behavioral patterns.
What are the differences between cybersecurity and fraud detection?
Cybersecurity and fraud detection are closely related disciplines, but they focus on different layers of risk and address different types of threats.
Cybersecurity is primarily concerned with protecting systems, networks, and data from unauthorised access and technical attacks. Its goal is to prevent breaches, malware, and intrusions by securing infrastructure through controls such as authentication, encryption, and network defenses. In essence, cybersecurity asks: “Is someone trying to break into or compromise the system?”
Fraud detection focuses on the behaviour taking place within the system, often after access has already been granted. Fraudsters frequently look like legitimate users, which means the challenge is not stopping entry, but determining whether actions such as transactions or account activity are genuine or abusive. Fraud detection relies on behavioural analysis, historical patterns, and machine learning models to assess risk. In short, cybersecurity protects the digital infrastructure, while fraud detection protects the business and customers from the misuse of that infrastructure
What types of fraud can these tools detect and in which sectors is mostly needed?
These tools are primarily used to detect payment fraud, account takeovers, and transactional abuse by identifying unusual patterns in user behavior rather than relying solely on fixed rules. In some cases, they can also help identify friendly fraud, which occurs when legitimate customers dispute valid transactions and can be particularly challenging to detect, for example, when a family member makes a purchase without the cardholder realizing it. They are especially valuable in sectors with high volumes of digital transactions, such as financial services, e-commerce, online platforms, and subscription-based businesses, where fraud evolves quickly and traditional detection methods struggle to keep pace.
Where do you see AI in the next coming years?
In the coming years, AI will move from being a supporting tool to a core decision-making layer across industries. We will see it embedded directly into products and operations, handling real-time decisions, personalisation, and risk management at scale. The biggest shift will be toward AI systems that are more transparent, regulated, and aligned with human oversight, especially in sensitive areas like finance, healthcare, and security.
How your services and products differentiate?
What differentiates our services and products is that we don’t offer isolated solutions, we build and operate a fully integrated fintech and cybersecurity ecosystem that covers the entire lifecycle of digital payments, risk management, and security.
At group level, our platforms span traditional finance, crypto payments, regulated PSP services, and cybersecurity, operating across multiple jurisdictions. This allows us to support businesses from payment acceptance all the way to fraud prevention, compliance, and secure infrastructure, under one unified technological vision.
Who is who
Fedros Avraam is the Chief Technology Officer of P.C ADMIN 22 LIMITED, a Cyprus-based fintech and software engineering company focused on building secure, scalable, and regulation-ready financial platforms. Fedros leads the company’s technology vision, with a strong emphasis on AI-driven fraud detection, payment systems, cloud architecture, and cybersecurity.
He has over 8+ years of professional experience in software development, fintech, and cloud infrastructure. His career began with hands-on engineering roles, where he worked on enterprise-grade systems and high-volume financial platforms. Over time, he progressed into senior and leadership positions, allowing him to combine deep technical expertise with strategic decision-making.