Build an explainable next-best-product recommendation system for banking on AWS

<p>Building a deep learning-based explainable next-best-product recommendation system helps banking institutions predict which product a customer needs next. Banks hold vast amounts of customer data, including transaction histories, product ownership records, demographic profiles, and behavioral patterns. Translating this data into actionable, personalized product recommendations remains a significant challenge. Traditional rule-based systems and collaborative filtering approaches often fail to capture the complex temporal patterns in customer product adoption journeys.</p> <p>In this post, we present the architecture and design decisions behind a Next-Best-Product (NBP) recommendation system using <a href="https://aws.amazon.com/sagemaker/ai/" target="_blank" rel="noopener">Amazon SageMaker AI</a> and PyTorch. We explain the reasoning behind a multi-tower neural network architecture, how learned attention provides per-customer explainability, and how AWS services work together to take this solution from research to production. This is an architectural overview, not a step-by-step deployment guide. Whether you are building recommendation systems for financial services or other domains with heterogeneous customer data, the architectural patterns described here can help you design more accurate and interpretable models.</p> <h2 id="prerequisites">Prerequisites</h2> <p>To follow along with the architectural patterns and code examples in this post, you need:</p> <ul> <li>An AWS account with permissions for SageMaker AI, <a href="https://aws.amazon.com/s3/" target="_blank" rel="noopener">Amazon Simple Storage Service (Amazon S3)</a>, <a href="https://aws.amazon.com/glue" target="_blank" rel="noopener">AWS Glue</a>, and <a href="https://aws.amazon.com/cloudwatch" target="_blank" rel="noopener">Amazon CloudWatch</a></li> <li>This solution requires an <a href="https://aws.amazon.com/iam/" target="_blank" rel="noopener">AWS Identity and Access Management (IAM)</a> execution role with access to the following AWS services. Create policies with least-privilege permissions scoped to only the resources needed for this solution.</li> <li><strong>SageMaker AI</strong> – Create, describe, start, stop, and delete training jobs, processing jobs, batch transform jobs, models, endpoints, endpoint configurations, pipelines, experiments, and monitoring schedules. InvokeEndpoint for real-time inference.</li> <li><strong>Amazon S3</strong> – Read and write access to the data bucket. Create and delete bucket. List, upload, download, and delete objects.</li> <li><strong>AWS Glue</strong> – Create, run, and delete ETL jobs. Create and delete crawlers. Create and delete Data Catalog databases and tables.</li> <li><strong>CloudWatch</strong> – Read and write access to log groups, log streams, and metrics. Delete log groups during cleanup.</li> <li><strong>IAM</strong> – Create and delete roles. Attach and detach policies. PassRole (restricted to the named execution role ARN, scoped to sagemaker.amazonaws.com and glue.amazonaws.com).</li> </ul> <p>For guidance on writing least-privilege IAM policies for SageMaker AI, see Identity-based policy examples for <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/security_iam_id-based-policy-examples.html" target="_blank" rel="noopener">SageMaker AI</a>.</p> <ul> <li>Familiarity with Python 3.11+ and <a href="https://docs.pytorch.org/docs/2.12/index.html" target="_blank" rel="noopener">PyTorch</a></li> <li>Required packages to create this recommender system: <ul> <li>Python 3.11+</li> <li>PyTorch 2.9+</li> <li>Pandas 2.3+</li> <li>NumPy 2.3+</li> <li>scikit-learn 1.7+</li> <li>Dask 2025.11+</li> </ul> </li> </ul> <blockquote> <p>We recommend using a virtual environment and scanning dependencies for known vulnerabilities using tools like pip-audit before deployment.</p> </blockquote> <ul> <li>Basic understanding of deep learning concepts (embeddings,