Khaled Dahmri

Développement et optimisation de la plateforme Content Management Expertise (CMX), une solution SaaS cloud-native et API-driven destinée à remplacer les plateformes legacy telles que Documentum, FileNet ou les NAS. Intégration de fonctionnalités avancées basées sur l’intelligence artificielle, notamment le tagging automatique, la recherche sémantique et la détection de la qualité du contenu.

Taches :

o Développement Python & outillage

o Refactorisation, stabilisation et industrialisation d’outils Python existants

o Conception et développement de nouveaux modules permettant la traduction de règles

métiers en workflows techniques automatisés.

o Participer à la mise en place de services de recherche sémantique et d’auto-tagging

intelligent

o Exploiter les relations entre concepts pour améliorer l’accessibilité et l’analyse des contenus

o Intégration de librairies avancées (NLP, IA générative, sémantique) dans les pipelines de

traitement documentaire.

o Réalisation de Proof-of-Concepts pour des fonctionnalités basées sur l’IA (tagging

automatique, recherche sémantique, détection de qualité de contenu).

o Optimisation de scripts Python

o Appliquer les bonnes pratiques de développement (PEP8, TDD, clean architecture)

o Recueil et formalisation des besoins auprès des experts métiers et des Product Managers.

o Traduire les logiques métiers formulés en langage naturel en règles automatisées

exécutables

• Amélioration de la qualité, robustesse et maintenabilité des outils Python.

• Contribution directe à la modernisation des systèmes de gestion documentaire et à la

migration cloud d’AXA.

Environnement technique : Python, GitHub, AWS, SQLite, Docker, Terraform, MarkLogic

BP2I is the IT department of BNP Paribas, responsible for implementing technological solutions to support internal operations, integrating data analysis and reporting tools, such as the Metrology Portal. Within the Metrology Portal team, composed of 4 members, my mission is to:

Implement processes to collect and process data from various sources such as Dynatrace, Nimsoft, and ServiceNow, regarding the BNP infrastructure (CPU, RAM, memory, file systems, network, incidents, etc.). This data is then analyzed and visualized in the form of Power BI dashboards (approximately twenty reports) and via search engines, accessible to all BNP Paribas businesses (application and infrastructure managers, managers, team leaders, etc.).

Developing PowerBI Dashboards

Propose and develop artificial intelligence models to meet the specific needs of different businesses.

Work performed:

Participation in the migration of old shell scripts to the Airflow platform and their deployment to production.

Creation and monitoring of Airflow workflows for data collection and processing.

Participation in the implementation of a report catalog (Power BI dashboards and JavaScript graphs).

Implementation and management of data quality processes (Data Quality Process) to ensure the reliability and accuracy of the information used in the dashboards and reports of the Metrology Portal, contributing to informed decision-making.

Development and testing of Deep Learning models (CNN) to assess the quality of the text in the resolution notes fields of incidents in ServiceNow

Technical environment: Python, Airflow, Hadoop, Hive, Pyspark, Mysql, PowerBI, Dynatrace, NimSoft, Visual studio code, Pycharm, Putty, Workbench, Gitlab, Gitlab CI/CD, TensorFlow, word2vec, Jupyter, Docker, FastAPI, IBM cloud.

The objective of this project is to set up a platform that allows the various SFR stores to identify customer profiles, calculate entries/exits, monitor customer activity in the store, and provide real-time reporting (visualization).

- Use a pre-trained model (YOLO) for person detection and track their trajectory.

- Implementation of a deep learning model for gender and age detection

- Implementation of a deep learning model for detecting different types of clothing.

- Implementation of a deep learning model for person recognition.

- Implementation of a deep learning model for mask detection.

- Transform the various models created into API mode and put them in a Docker image.

- Prepare the backend to retrieve video streams from different cameras in order to analyze them frame by frame in order to detect people, their ages and genders, and calculate their entries/exits and store this information in an Elasticsearch database.

- Prepare the frontend to display the results of the analysis in real time.

- Prepare the reporting in graph format in Kibana and integrate it into the frontend.

- Prepare a Docker image for the frontend and backend.

Technical environment: Python3.8 (keras, tensorflow, pytorch, Django, opencv), elasticsearch, kibana, gitlab/CI, Docker, Pycharm, Visual Studio Code.