Matthieu Soumbo

- Piloting the design of an aeronautical maintenance digital service, structuring complex data flows (ingestion, analysis, restitution) in an AWS cloud environment, with cross-functional coordination of business and technical teams.

- Definition and prioritization of the product backlog

Writing detailed functional specifications (SFD) and user stories with acceptance criteria

- Design of user journeys and interfaces (fleet visualization, alert analysis, event management, recommendations tracking).

Product piloting in an AWS cloud environment (data platform, ingestion, analytics), in coordination with technical teams.

I monitored data quality in real-time, enabling teams to make informed decisions at each sprint. Continuous improvement was at the heart of the project, with regular feedback from stakeholders to refine indicators and adjust reports according to business needs.

I worked closely with cross-functional teams (technical, business, and data) in an agile approach to define and iterate on data governance processes. Scrum ceremonies, such as daily scrums, sprint retrospectives, and sprint reviews, facilitated team synchronization and rapid adjustment of priorities based on emerging needs.

Through a series of sprints dedicated to evaluating the perception of data quality, I deployed targeted surveys that provided actionable feedback. These results were used to refine our data quality indicators and adjust processes in real-time, ensuring that improvements were directly aligned with end-user needs.

I developed and rapidly delivered functional prototypes in short cycles (sprints), integrating reporting and monitoring tools. These prototypes were tested and validated with business teams before being integrated into the production environment, enabling faster adoption and increased return on investment.

I presented 6 solutions to business problems using data modeling and visualization on concrete projects validated by industry experts.

- Statistical analysis and machine learning. Development of predictive algorithms capable of identifying anomalies and providing recommendations to improve public health on a large scale.

- Exploratory data analysis and data set preparation for modeling, design of machine learning models (regression, neural networks, etc.), and performance optimization through hyperparameter selection. The goal was to optimize the accuracy of energy forecasts and reduce potential energy costs through better resource management.

- Customer profile detection through unsupervised learning. Use of learning algorithms (clustering) to identify distinct customer segments based on their purchasing behavior. Implementation of a feature engineering process to improve model performance.

- Design of a convolutional neural network (CNN) for visual feature extraction. Deployment of the model on photos, then adaptation for video processing with real-time detection.

- Development of an NLP-based model to analyze questions and suggest appropriate tags. Feature engineering to improve the accuracy of the suggestion system, search, and organization of information for users.