The existing credit risk assessment process relied on expert judgment, which, while effective, limited the company’s ability to scale. Each application required thorough analysis by human experts, creating a bottleneck in processing speed and efficiency. As the demand for factoring services grew, the need for a more agile and scalable system became imperative. The company sought a solution that would uphold its established business principles while leveraging machine learning to automate and enhance decision-making capabilities. The goal was to create a system that operates continuously, ensures objectivity in risk evaluation, and significantly expands the company’s ability to process applications.

The project began with an in-depth analysis of the company’s existing risk assessment processes to ensure alignment with business objectives. The solution included the construction of two sets of machine learning models tailored to the client’s needs. The first was a black-box model with an explainable AI (XAI) mechanism, providing insights into the decision-making process. The second was a fully transparent model, allowing for greater interpretability and alignment with regulatory and operational requirements. An interactive application was developed to allow the client to compare both model types and select the most suitable one. Additionally, a dedicated application was built to facilitate the seamless integration and operational use of the chosen machine learning model, ensuring smooth deployment within the company’s workflow.

The implementation of machine learning models transformed the factoring application process, enabling automation and objective risk assessment. The new system operates in a 24/7 model, eliminating previous time constraints and increasing efficiency. The company can now process a significantly larger number of applications within a short period, allowing for rapid business expansion. By integrating AI-driven decision-making, the company has enhanced its ability to evaluate risk consistently while maintaining compliance with its established business principles. The scalability of the new system ensures long-term growth opportunities without additional strain on human resources.

In-depth analysis of risk assessment processes and design of the solution to fit the company’s application flows. Development of two sets of ML models: black-box models (with an XAI explainability mechanism) and transparent models. Preparation of a report and an interactive application enabling the client to decide on the preferred model type. Development of an application allowing the use of the selected type of machine learning models.

The development process involved a combination of advanced machine learning techniques tailored to credit risk assessment. The black-box model utilized deep learning algorithms enhanced with explainable AI mechanisms, allowing for interpretability while maintaining predictive power. The transparent model was built using interpretable statistical and machine learning methods to align with regulatory expectations. The model selection application employed a comparative analysis framework, enabling the client to assess key performance metrics and make an informed decision. Continuous testing and validation ensured that the models met both business and regulatory requirements, delivering a reliable and scalable solution for factoring risk assessment.