À propos de Visar
Data executive with 12+ years of strategic leadership in data architecture, engineering,
governance, and innovation, driving enterprise-wide data initiatives that fuel business
growth and operational excellence.
Anglais
Bilingue ou natif
Français
Capacité professionnelle complète
Turc
Bilingue ou natif
Albanais
Bilingue ou natif
Accepte de travailler sur site
Paris (jusqu’à 50 km)
Expériences
- Trace OneLead Data ArchitectEDITION DE LOGICIELSfévrier 2026 - Aujourd'hui (6 mois)Paris, FranceContextDesigned, architected, and implemented a modern GCP-based data platform to migrate Trace One from legacy SQL Server infrastructure to a scalable, cloud-native analytics stack.Key Challenges-Multi-tenant and multi-environment architecture with strict POC budget constraints.-Dynamic product lifecycle data with frequently changing product specifications and schema evolution.-Complex stakeholder alignment across Data, DataOps, DevOps, Infrastructure, and Security leadership.-Need for reusable pipelines deployable both in cloud and client-owned on-premise environments.Solution-Defined the target data architecture using BigQuery as the warehouse and processing engine, dbt for transformations, Composer/Airflow for orchestration, and Terraform for infrastructure provisioning.-Designed dynamic Airflow DAGs to support multi-tenant ingestion, configurable APIs, and customizable entity onboarding.-Built a three-layer medallion architecture: Bronze for landing, Silver for self-service warehousing, and Gold for reporting and data marts.-Used dbt macros to handle schema changes dynamically at the warehouse layer.-Implemented idempotent ingestion and transformation logic through primary key definitions and dbt merge strategies.-Chose Airflow and dbt to ensure pipeline portability for private-cloud and on-premise client deployments.Impact- Secured alignment and approval from all key technical and business stakeholders.-Delivered an operational multi-tenant data platform supporting configurable ingestion and self-service warehousing.-Enabled onboarding of the internal data team onto a scalable, maintainable, and reusable data platform.
- Crédit Agricole Alpes ProvenceLead Data ArchtiectBANQUE & ASSURANCESavril 2025 - novembre 2025 (7 mois)ContextPropulse, a Crédit Agricole solution for entrepreneurs, relied on NoSQL application databases without a centralized data platform. This created data quality issues, inconsistent reporting, customer duplication, and unreliable conversion attribution.Key Challenges- Build a modern data lake while aligning with the company’s existing AWS stack and team skillset.- Resolve complex customer consolidation issues involving duplicate, merged, and split customer identities across inconsistent PII.- Create a trusted foundation for conversion tracking and attribution modeling.- Protect sensitive PII and prevent exposure within the data lake.Solution- Designed and implemented an AWS-based data platform using S3, Redshift Serverless, Lambda, Step Functions, EventBridge, and AWS CDK.- Built a two-layer medallion architecture with Landing and Warehouse zones.- Standardized ingestion through Lambda functions loading JSON structures from S3 into Redshift landing tables.- Kept transformation logic in SQL and Redshift stored procedures to reduce onboarding complexity for the internal team.- Developed a robust 360° customer view using multi-layer SQL clustering and ranking logic to identify and resolve customer merge/split scenarios.- Encrypted PII in the landing layer to prevent developer exposure to sensitive data.Impact- Delivered a fully operational, autonomous AWS based data platform.- Improved customer consolidation by 20%, addressing legacy gaps where roughly 30% of customers were previously unconsolidated.- Increased identified conversions by 30% by uncovering hidden conversions lost due to poor legacy data quality.- Established a reliable single source of truth for activation, conversion, and attribution use cases.
- OppScienceEngineering ManagerSECTEUR PUBLIC & COLLECTIVITÉSseptembre 2024 - Aujourd'hui (1 an et 11 mois)Paris, France- Defined technical and product roadmap for semantic AI platform (Spectra) and led a cross expertise 7 member team to 3 consecutive quarterly releases for law enforcement software. (Java, Python, fine tunable ML, K8s, semantic, graph)- Launched Semantic Studio Standalone, a standalone app for semantic extraction creating new sales opportunities.- Delivered on-premise fine tunable Flair relation extraction model, boosting investigative accuracy.- Delivered gemma:3n LLM model for on-premise summary generation.
Avis
Recommandations
Soyez le premier à recommander Visar
Contribuez à la réussite de ce freelance en partageant votre expérience de collaboration avec lui.
Ces profils de freelance correspondent également à vos critères
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Formations
- Master of ScienceEcole Centrale Paris2016Master of Science in Information Technologies for Business Intelligence Ecole Centrale Paris, France - Decision Science, Data Mining, Visual Analytics (Tableau, R) Université Francois-Rabelais de Tours, France - Data Mining, Information Retrieval, Data Warehousing Université libre de Bruxelles, Belgium - Data Warehousing, Business Process Management