Anis Tlilane

As part of the DDF (Fraud Detection) project, I contributed to the design and deployment of an intelligent and centralized fraud detection platform, hosted on Palantir Foundry, covering several risk typologies (SF Score, Fraud, LFT...) and deployed by business unit. I led the implementation of the Leasing & Factoring scope, ensuring the complete data value chain: from ingestion to scoring, fraud scenarios, and operational alert management.

Responsibilities:

• Business needs framing, data source identification, and business rules definition for transformation and enrichment.

• Implementation of data collection and ingestion processes from AWS S3 (workflow automation, heterogeneous format management, high volume).

• Development of processing pipelines using PySpark/Python: unzip, parsing, cleaning, normalization, deduplication, and quality controls.

• Design of the target data model and object modeling in Ontology (Palantir service): definition of entities, relationships, and cardinalities.

• Implementation of data mart population via Ontology and setup of indexing for optimized search and analysis.

• Development of a semantic search engine for individual scoring and identification of risky links.

• Design and configuration of fraud scenarios (rules, aggregations, alert thresholds).

• Implementation of the Case Manager for alert lifecycle management (generation, viewed scored profiles, investigation workflow).

• Implementation of processing scheduling and planning (dependency management, scheduling, and monitoring).

• Implementation of data lineage ensuring complete data traceability, from source to generated indicators and alerts.

Project: ACPR

As part of the prudential supervision of the banking and insurance sector, the Autorité de Contrôle Prudentiel et de Résolution (ACPR) conducted an audit on the Generali group. An internal project was initiated to effectively respond to this regulatory requirement by providing the expected data marts and indicators.

Responsibilities:

• Extraction, collection, and ingestion of raw data into the Data Lake.

• Normalization, cleaning, and validation of data from multiple sources.

• Cross-analysis of similar sources to identify correct data and complete it.

• Construction of data marts through automated processing.

• Optimal coordination of interdependencies between processes.

• Structuring of the project into business Flow Zones.

• Automation of workflows and scenarios using Dataiku.

• Design of dashboards for KPI monitoring.

• Proactive detection and correction of functional anomalies.

• Contribution to business analysis and strategic decisions.

Project: Data Product

Following the 2023 ACPR audit revealing weaknesses in Financial Security steering and data structuring, a Data Product project based on a data mesh architecture was launched in 2024 to optimize data quality and utilization.

Responsibilities:

• Data preparation, enrichment, and segmentation.

• Implementation of processing for Data Product creation.

• Establishment of dedicated data quality processes.

• Implementation of data and processing monitoring.

• User access and service account management on Snowflake.

• Secure management of secrets and configuration variabilization.

• Automated deployment via a CI/CD pipeline.

• Creation of DAGs for orchestration and execution planning.

• Development of a GitFlow and dissemination of best practices.

• Execution of PoCs to prepare for technical migration.

Project: Data Platform Exploration

Design and deployment of an industrialized DataLab platform on AWS dedicated to the complete Data Science cycle (model creation, training, and deployment), data management (catalog, privacy, quality), and business reporting.

Responsibilities:

• Creation of the ML platform in Infrastructure as Code.

• Creation of custom SageMaker environments with Git, IAM, S3 management, and specific Docker images.

• Secure administration of S3 buckets (least privilege, IAM policies).

• Feature centralization via SageMaker Feature Store.

• Automation of the model cycle (preparation, training, deployment) via SageMaker Pipelines.

• Model management and retraining via SageMaker Model Registry.

• Implementation of Amazon RDS databases connected to Power BI by business unit.

• Construction and exposure of the Data Catalog.

• Processing and ingestion of Power BI logs and SAS data.

• Technical assistance and training for Data Science teams.

• Corrective and evolutionary maintenance of the platform.

Project: Data Platform Engine

Design and management of a scalable and secure AWS Data Platform for real-time (streaming) and batch data processing, ingestion, and storage, ensuring the reliability and performance of the data flows feeding the Data Lake.

Responsibilities:

• Development of real-time streaming processing (MSK, EMR, Spark Streaming, Scala).

• Implementation of batch ingestion processes (Step Function, Lambda, Python).

• Secure cross-account data transfer via S3 and IAM.

• Construction of Consumer Zones (data marts) on RDS.

• Kafka administration via Amazon MSK (retention, resource sizing).

• Real-time monitoring of performance and scalability via Grafana.

• Securing access to AWS resources via SSM (bastion removal).

• Corrective and evolutionary support (purging, data transfer, processing restarts).