fr
en
Deploy AI and machine learning in FinTech services
AI applications in finance must connect model relevance with business requirements. Explore possible applications, implementation conditions and limitations. Strengthen your ability to engage with technical teams and assess results critically.
- Duration
- 3 days 21 hours
- Code
- FN003FR Code
Presentation
In financial technology (FinTech), real-time transaction analysis is no longer optional: it is essential to survive fraud and regulatory pressures. This intensive 3-day course takes you beyond static rules to implement a genuine artificial intelligence strategy that secures operations and optimises customer scoring.
Explore machine learning algorithms, including random forests, isolation forest and gradient boosting, specifically adapted to imbalanced financial data. Learn to build reproducible pipelines, clean complex datasets and ensure model explainability for compliance audits covering AML/CFT and GDPR.
By the end of the programme, you will not just build models: you will know how to put them into production. Master MLOps practices to deploy solutions through APIs, monitor data drift and automate production retraining, ensuring the long-term reliability of your financial information systems.
Objectives
By the end of this AI and machine learning course, you will be able to:
- apply AI techniques to FinTech-specific challenges, including scoring, KYC and fraud;
- design and train effective anomaly and fraud detection models;
- deploy a machine learning model to production through APIs;
- master MLOps best practices for model monitoring and maintenance;
- ensure algorithm explainability to meet regulatory constraints.
Program
Module 1: Learning ML fundamentals and understanding FinTech
- An overview of AI in finance and analysis of trends in Africa and worldwide.
- Mapping key use cases: credit scoring, KYC compliance and transaction optimisation.
- Integrating GDPR, BCEAO and AML/CFT regulatory constraints from the design stage.
- Fundamentals of supervised and unsupervised machine learning and the CRISP-DM methodology.
- Advanced financial data preprocessing: cleaning, feature engineering and class imbalance management.
- Classification and detection algorithms: logistic regression, random forests and gradient boosting.
Hands-on exercises
- Map relevant AI use cases for your company.
- Explore a transaction dataset with Python, Pandas and Matplotlib.
- Preprocess and clean simulated real-world data.
- Develop initial detection models with scikit-learn.
Module 2: Building advanced fraud and anomaly detection models
- Selecting discriminative features and managing streaming time-series data.
- Evaluating models using fraud-appropriate metrics: precision, recall, F1-score and AUC-ROC.
- Implementing unsupervised methods, including Isolation Forest and One-Class SVM, to detect contextual anomalies.
- Creating reproducible ML pipelines with scikit-learn to automate processing.
- Analysing model interpretability with SHAP and LIME to justify decisions to auditors.
Hands-on exercises
- Train a fraud detection model and optimise its hyperparameters.
- Implement an Isolation Forest model on a simulated data stream.
- Build a complete detection pipeline combining preprocessing and a model.
- Interpret detection model results with SHAP or LIME.
Module 3: Putting models into production and maintaining them with MLOps
- MLOps foundations and CI/CD pipeline integration for automation.
- Deploying the model as a REST API using FastAPI or Flask on a server or in the cloud.
- Managing data drift and establishing automatic retraining strategies.
- Monitoring performance and using tracking tools such as MLflow.
Hands-on exercises and final workshop
- Set up model tracking with MLflow.
- Deploy a model through an API in a simulated cloud environment.
- Detect drift in a transaction stream.
- Complete a team project from raw dataset to deployed model.
- Transform a raw dataset into a deployed, integrated model in the final workshop.
Audience
This course is intended for technical and analytical professionals in finance, including:
- data scientists and data engineers designing analytical engines to improve decision reliability;
- back-end developers integrating predictive models into existing application architectures;
- business analysts using data to detect fraudulent behaviour;
- Product Owners and Product Managers leading innovation and FinTech solution roadmaps.
Prerequisites
The following prerequisites apply:
- Professional experience: experience in data analysis, development or technical product management is recommended.
- Basic knowledge:
- proficiency in Python fundamentals for data manipulation;
- elementary mathematics and statistics;
- general understanding of financial services.
Teaching and assessment methods
- Initial skills assessment
- Training materials provided to participants
- Continuous assessment throughout the course
- End-of-course feedback questionnaire
- Combination of theory and practical application
- Attendance records
- Post-course follow-up evaluation
- Practical exercises
Course highlights
- Strong sector focus: work on real FinTech use cases involving scoring, fraud and KYC, integrating specific BCEAO and AML/CFT constraints for compliant solutions.
- Production-ready methodology: go beyond theory to master the MLOps chain, from Pandas data preparation to API deployment with FastAPI and MLflow.
- Advanced detection techniques: use specialist algorithms such as Isolation Forest and autoencoders, and manage the highly imbalanced datasets typical of fraud.
- Complete end-to-end project: complete a final workshop from scratch, from raw dataset to deployed model, preparing you to work immediately on your own projects.
Dates and sessions
Choose the date and delivery format that suit you.
No upcoming sessions are currently available.
Brand names and logos mentioned in this course description, such as Python, Pandas, Scikit-learn, MLflow, AWS and Azure, belong to their respective owners. Their use for educational purposes does not constitute a commitment or partnership.