Othman FAIQ

– Support GPU-based infrastructure to scale and industrialize AI/ML initiatives across modern, resilient platforms, both on-premises and in the cloud – Design and maintain AI/ML pipelines, including AI agents and Retrieval-Augmented Generation (RAG) solutions

Project: Big data application that provides scenarios of pricing parameters schocks. the scenarios will then be used to compute the main risk measures of Natixis like VaR, Stress Tests or Sensitivities Key Contributions: – Implemented a daily data ingestion workflow for market data referentials, triggered by a Kafka-based event system – Developed risk model calculators aligned with specifications from Risk Management Officers – Built and maintained sanity checks to ensure data completeness and integrity – Designed and Implemented a fallback mechanism that reuses previous day's data when sanity checks detect missing or incomplete inputs – Managed the Jira board and contributed to PI planning in an Agile environment

Project: Integration of a UEBA solution (User and Entity Behavior Analytics) for e-communications surveillance (emails, chats, and other messaging systems) Key Contributions: – Developed custom parsers and implemented data ingestion pipelines for multiple sources: Skype, Symphony, Bloomberg, Reuters, ICE, etc. – Built NLP-based tools for detecting duplicates, near duplicates, and disclaimers across communication content – Coordinated with business teams to define and implement alert detection rules (lexicons, filters) and support User Acceptance Testing (UAT) – Maintained and monitored the ElasticSearch infrastructure used for storing and querying surveillance data – Acted as technical liaison with the vendor for the successful migration to a new version of the surveillance platform

Project: Development of Data Quality Indicators for a datalake-based platform centralizing Trade and Pre-Trade data within the Corporate & Investment Banking perimeter Ket Contributions: – Developed and implemented Machine Learning models to detect anomalies in the volume of data ingested by the platform (e.g., missing data, spikes, unusual patterns) – Industrialized and scaled the models to operate efficiently in a datalake environment

– Building a NLP (Natural Language Processing) application for text data analysis – Text classification and topic modeling