The latest articles on DEV Community by Rafael Contente (@rafaelcontente). https://dev.to/rafaelcontente https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4015533%2F491ba90e-7602-4c28-a2c9-88d16d9510e8.png https://dev.to/rafaelcontente en Rafael Contente Sat, 04 Jul 2026 21:06:09 +0000 https://dev.to/rafaelcontente/building-trustworthy-ai-beyond-prediction-527o https://dev.to/rafaelcontente/building-trustworthy-ai-beyond-prediction-527o <p>Building Trustworthy AI Beyond Prediction</p> <p>Over the past months, I've been developing CREDERE, an AI engineering project exploring a question that I believe is becoming increasingly important:<br> What happens after a machine learning model makes a prediction?<br> Most discussions around AI still focus on model accuracy.<br> In regulated domains like credit scoring, however, accuracy is only one part of the problem.</p> <p>A production system also needs to answer questions such as:<br> Is the decision legally compliant?<br> Can it be explained without hallucinations?<br> Can every explanation be verified?<br> Can a human intervene when necessary?<br> Can every decision be audited afterwards?</p> <p>Those questions cannot be solved by a better classifier alone.<br> They require software architecture.</p> <p>Instead of building another predictive model, I designed a neuro-symbolic architecture where statistical learning is complemented by deterministic reasoning.</p> <p>The public repository currently focuses on two production-oriented modules:<br> Compliance Engine — deterministic regulatory rules capable of overriding model predictions when legal constraints are violated.</p> <p>Explanation Engine — customer-facing explanations generated without factual hallucinations through structured templates and independent fact verification.<br> Today I'm also publishing the CASE_STUDY.md, which documents the engineering journey behind the project.</p> <p>Rather than describing only what was built, it explains:</p> <p>the original problem;<br> the architectural decisions;<br> alternative approaches that were considered;<br> engineering trade-offs;<br> validation strategy;<br> lessons learned;<br> limitations;<br> future directions.</p> <p>I believe engineering documentation is often overlooked, yet it is one of the clearest indicators of how a system was actually designed and why certain decisions were made.</p> <p>If you're interested in:</p> <h1> Trustworthy AI </h1> <h1> Explainable AI (XAI) </h1> <h1> Neuro-Symbolic AI </h1> <h1> AI Governance </h1> <h1> Software Architecture </h1> <h1> AI Engineering </h1> <h1> Credit Scoring </h1> <h1> RegTech </h1> <p>The project can be viewed on my GitHub: <a href="https://github.com/rafaelcontente/CREDERE" rel="noopener noreferrer">https://github.com/rafaelcontente/CREDERE</a></p> <p>I'd genuinely appreciate your feedback.</p> <p>I hope the case study proves useful not only for understanding CREDERE but also as a practical example of documenting AI systems beyond predictive performance.</p> <h1> ArtificialIntelligence #AIEngineering #SoftwareArchitecture #MachineLearning #ExplainableAI #TrustworthyAI #ResponsibleAI #NeuroSymbolicAI #CreditScoring #RegTech #OpenSource #Python #EngineeringDocumentation </h1> ai fintech machinelearning softwareengineering