Unique Anti-fraud tools project – PIN-UP.TECH developers change the approach to fighting fraud

The Anti-fraud tools project is an example of how the team has revolutionized the approach to combating fraud in the online gambling market. Spoiler alert: the PIN-UP development will be available worldwide in the near future.

How it all started

PIN-UP has always focused on automating as many risk management and anti-fraud processes as possible. The first tools were implemented as part of the PIN-UP core product, which allowed closing the key problem areas at the top level. Next, they worked with a third-party anti-fraud platform. However, PIN-UP's growth rate is so rapid that the extensive system of tools on various ecosystem products ceased to meet all needs at some point. Another and very important motivation for change is the immense number of ideas that have been accumulated, which were impossible to implement under previous conditions. Therefore, it was only a matter of time before the PIN-UP.TECH created its own centralized anti-fraud platform.

Initially, the main focus of development was the RnD service fingerprint. Analyzing a user's device is a very important component of anti-fraud. PIN-UP has developed its own proprietary algorithm, which they are now fully satisfied with. At the same time, the product team was carefully and meticulously selected: a project manager, developers, testers, DevOps, and front-end engineers.

Over the year, PIN-UP.TECH has built a powerful Anti-fraud department of more than 10 specialists, where everyone is in their place: they do not just fulfill their tasks, but come up with specific ideas and solutions. Initiative is a hallmark of this team. Everyone has a say and opportunities to realize their potential. This is largely due to the effective management of Project Manager Vitalii Voitsekhivskyi. 

The alpha release took place in June 2023, but the team faced many challenges long before that. At the start of the project, it was necessary to think through in detail all the scenarios of AFT's interaction with the PIN-UP platform, the requirements for microservice architecture, and assess the load. And only then all business needs had to be translated into technical language for the team. 

It was important for us to build the project architecture looking ahead to the B2B segment so that we wouldn't have to rewrite thousands of lines of code later. We have an ambitious goal — to cover all the needs of PIN-UP and enable other businesses to use our solution.
Why have we developed our own anti-fraud product? There are absolutely no alternatives with similar capabilities 😊 Our anti-fraud tools implement ideas that are not yet on the market. 

We have now passed the first challenging stage — the MVP implementation. We tested different architecture options, different combinations of databases and data management approaches. Having completed the load testing, we decided on the final system configuration. What drives us further? Confidence that we will seamlessly centralize all anti-fraud tools in one product and start bringing business value from the first day of AFT integration into the PIN-UP product. - Volodymyr Todurov, Chief Analytics Officer

What makes the project unique

The Anti-fraud tools project by PIN-UP.TECH combines traditional anti-fraud rules with advanced technologies such as machine learning to achieve the highest efficiency and reliability. 

The user device profiling module can collect more than 300 different parameters both in a web browser and from native applications (Android/iOS). A certain set of such parameters can be used to build a unique device fingerprint: deviceID — as a browser-independent and browserID — as a browser-dependent anonymous identifier. - Oleg Potapov, RnD engineer

• AI (artificial intelligence) helps them assess the level of trust in a transaction. All requests go through standardized checks, after which the result is sent to the AI service, where the same transaction is evaluated by AI, and the response is adjusted. In addition, if a transaction is marked as suspicious after passing the checks, this is taken into account by AI models for further analysis. As a result, the team has a broad response on various metrics whether the specified transaction can be trusted or not.
• Event-driven system ensures that all notifications are asynchronous. Therefore, the system will not slow down while waiting to be processed by third-party services, and the response will be provided immediately upon receipt.
• The microservice architecture of the product allows you to easily and, most importantly, quickly scale up as the system load increases.
• Security. Their architecture takes into account security and personal data protection aspects — information about users is depersonalized and stored in encrypted form in databases such as Aerospike and ArangoDB.

This tool contributes to the increase of users' trust in gambling platforms and their activity. The flip side of the coin works here. By conducting scoring, specialists identify both potential fraud and potential loyal users and VIP customers. Thus, by segmenting customers from the first click on the website, they can provide VIP customers with better service: higher limits, faster KYC checks, better bonuses, and instant payouts. Of course, information about both VIP and other customers is securely protected and stored in encrypted form.

An incredible amount of attention has also been paid to data analysis, for example, the module of customized dashboards. Now there is no need for external BI tools — you can easily set up a convenient configuration of dashboards and monitor the target metrics within the platform. 

You no longer need to export different tables and dig through the data manually to understand the scope of the fraud case. It takes 5-7 clicks and all connections in the system are visualized. 

The main task of this development is to meet the needs of risk management, in particular in the online gambling segment. AFT covers all areas of business operations: payment systems, bonus campaigns, loyalty programs, affiliate programs, game styles, external attacks on the product and internal vulnerabilities, attempts to account takeovers, social engineering, etc. The expertise of the PIN-UP.TECH team allows them to design all possible scenarios.

Machine learning helps achieve greater accuracy in fraud prevention at different stages of customer contact. On their own, 7 arbitrary user transactions (registration, deposit, play, log-out, log-in, bonus activation, withdrawal) may not look risky. But if you build a portrait of user behavior — in what sequence the transactions were made, at what time of day, and how fast the transactions were processed through the system — it can give an unexpected result. Segmenting the user journey by behavioral patterns, they build a second line of defense in case each transaction did not look potentially risky individually.

Key components: architectural essence of Anti-fraud tools

In the world of innovation and development, the success of a product is inextricably linked to its technical architecture — every line of Anti-fraud tools code, every component, and every technological choice has its own unique role in the fight against online gambling fraud. Amazing teamwork of the developers allowed them to create a product that not only works perfectly, but also has a strong potential for further development. 

The structure of Anti-fraud tools includes, among other things, such components as scoring, fingerprint, and backoffice. They interact via microservices and NATS. Scoring performs predefined checks, fingerprint enriches transactions with data, and backoffice allows you to manage rules and view transaction history.

For the development and implementation of the product, we used a microservice approach with a focus on Event-Driving Architecture based on NATS, which allows us to distribute the load on the system and be as efficient as possible in terms of query execution speed. One of the most important tools for us is Aerospike and Redshift databases, which helped speed up the system's operation by several times. In addition, ArangoDB has become an indispensable tool for finding relationships between entities, as it helps to effectively determine whether a transaction is related by any attribute to fraudulent transactions already detected. - Maksym Tkach, Tech Lead

Plans and prospects

The PIN-UP.TECH team plans to integrate its product within the PIN-UP Global ecosystem and subsequently offer it to external customers. The next steps involve continuous improvement and expansion of the project's capabilities. 

In fact, we are now consolidating all anti-fraud and risk management tools into a single platform. Step by step, we are adding new scenarios for both B2C and B2B interactions within the ecosystem. This is a powerful tool that guarantees all stakeholders a good night's sleep, as they will know that a powerful system protects the main aspects of business operations from fraud. Anti-fraud tools are not only our request, but also the market's request, so we are confident that the product will be warmly welcomed by potential B2B customers. And we will speak the same language with them, having end-to-end expertise in igaming. - Volodymyr Todurov, Chief Analytics Officer

While working on the new product at an intensive pace, PIN-UP.TECH does not forget about the importance of expanding the team and strengthening expertise. In the near future, the team will be looking for experienced developers, analysts, and specialists in machine learning.

Conclusion

Volodymyr Todurov's development team has proved that innovation is a prerequisite for the success of any project. The world is evolving rapidly, and if you are not in the trend, you fade away. The introduction of new technologies, such as AI, increases productivity and opens up new opportunities for businesses, and smart management and proactivity of the team allow them to be the first to implement the best solutions.