Omar ABDEDDAIM

Embedded within EDF's Data Management Service, I contribute to building intelligent, production-grade AI and data solutions in a complex hybrid industrial environment, working closely with platform, security, and business teams to deliver ML-powered use cases and automated analytics workflows. – Developed and deployed ML models for predictive analytics on industrial asset data, enabling proactive maintenance and operational optimization – Built time series forecasting pipelines using Python and PySpark to anticipate energy consumption and production anomalies – Designed automated reporting workflows that replaced manual analysis processes, reducing reporting cycle time by over 60% – Built RAG-based chatbot solutions using LangChain integrated with Pinecone (vector database) and internal EDF knowledge bases, leveraging OpenAI and Mistral AI models – Designed the architecture of an Agentic AI platform using Agent-to-Agent (A2A) communication and Model Context Protocol (MCP) for orchestrating multi-agent data quality workflows – Implemented semantic search capabilities on top of structured and unstructured data assets to accelerate expert knowledge retrieval – Implemented scalable data pipelines using Airflow and Databricks, processing multi-source industrial datasets across on-premise and cloud environments (AWS, GCP) – Designed and deployed distributed data processing jobs with PySpark for large-scale ETL/ELT operations – Structured Data Lake zones (ingestion, processing, exposure) aligned with data sensitivity and governance requirements – Integrated multi-source data streams using Kafka (streaming & batch ingestion) and Pub/Sub – Deployed and tracked ML models using MLflow (experiment tracking, model registry, model serving) – Containerized and deployed ML services using Docker and Kubernetes (GKE), ensuring reproducibility and scalability – Implemented CI/CD pipelines for automated model retraining and deployment – Mentored junior data and ML engineers on LLM integration patterns, MLOps best practices, and Agentic AI design – Led internal knowledge-sharing sessions on RAG architecture, vector databases, and prompt engineering

Led the design and deployment of ML-powered analytics solutions across a multi-entity hospitality group, transforming raw operational data into intelligent, automated decision-support systems. – Developed and deployed ML models for demand forecasting and revenue optimization, predicting hotel occupancy rates and pricing dynamics across multiple business entities – Built fraud detection pipelines on transactional data using supervised learning models (XGBoost, Random Forest) deployed via FastAPI REST endpoints – Implemented time series forecasting for seasonal trend analysis and operational planning using Prophet and LSTM-based models – Designed a RAG-based internal assistant using LangChain and Hugging Face LLMs to enable semantic search over operational documentation and procurement knowledge bases – Integrated vector search (Pinecone) with structured data sources to power hybrid retrieval pipelines for business users – Prototyped an Agentic AI workflow for automated anomaly reporting, combining LangGraph agents with data quality triggers – Built and industrialized data pipelines using PySpark and Airflow for multi-source ingestion (booking systems, CRM, financial platforms) across AWS and GCP – Designed ETL/ELT workflows on AWS (S3, Lambda, EC2) and GCP (GKE, Cloud Run) with data quality and monitoring mechanisms – Structured Data Lake architecture aligned with data governance and business consumption needs – Implemented batch and streaming data integration for real-time and analytical processing – Deployed ML models in production using MLflow and Docker, with automated retraining pipelines triggered by data drift detection – Managed containerized deployments on GKE and Cloud Run, ensuring scalability and fault tolerance – Implemented CI/CD pipelines for model versioning, testing, and deployment automation – Designed and executed A/B testing frameworks for pricing strategy validation and product recommendation optimization – Automated operational and management reporting using Streamlit dashboards, replacing manual Excel-based workflows – Contributed to defining the target architecture of the data platform, considering hosting, security, and performance constraints – Implemented data governance frameworks (roles, responsibilities, access rules, usage guidelines) – Structured data catalogs and naming conventions for cross-functional data discovery – Wrote technical documentation (Architecture Design Documents, data flow diagrams, data models)

Contributed to a large-scale data modernization program using a Data Mesh approach, building ML-ready data infrastructure and intelligent analytics pipelines across multiple business domains. – Developed predictive models for supply chain optimization and product demand forecasting, reducing overstock and stockout incidents across distribution networks – Built automated anomaly detection pipelines on product quality data using isolation forest and statistical process control methods – Designed feature engineering workflows using PySpark on AWS EMR, preparing large-scale datasets for ML consumption – Implemented scalable batch ingestion and transformation pipelines using Python, PySpark, and Airflow (MWAA) on AWS – Structured Data Lake zones (raw, curated, consumption) aligned with Databricks Delta Lake architecture patterns for reliable ML data access – Integrated multi-source data (ERP, CRM, IoT sensors) into a unified analytical platform with end-to-end lineage tracking – Contributed to defining the target data architecture (Data Lake zones, ingestion flows, domain-based data exposure) – Built and industrialized data pipelines using PySpark and Airflow for multi-source ingestion (booking systems, CRM, financial platforms) across AWS and GCP – Designed ETL/ELT workflows on AWS (S3, Lambda, EC2) and GCP (GKE, Cloud Run) with data quality and monitoring mechanisms – Structured Data Lake architecture aligned with data governance and business consumption needs – Implemented batch and streaming data integration for real-time and analytical processing – Defined and implemented data governance rules (quality, traceability, responsibilities) – Set up data quality monitoring and flow supervision mechanisms ensuring ML model input reliability – Structured naming conventions and data catalogs for cross-domain data discovery – Managed data access and permissions by domain (Data Mesh logic) – Automated reporting and analytics workflows using Python scripts and AWS Step Functions, replacing manual business reporting processes – Implemented CI/CD pipelines for data pipeline deployment and version control

As ML Lead and technical referent at IBM's CIC, I delivered AI and data engineering solutions for major international clients (British Petroleum, Humana, CVS Pharmacy) across energy, insurance, and healthcare – progressively growing from Data Engineer to Senior ML Lead. – Designed and deployed fraud detection models for insurance and healthcare clients (Humana, CVS Pharmacy) using supervised classification algorithms on transactional and claims data – Built predictive analytics pipelines for energy consumption forecasting at British Petroleum, integrating sensor data with historical billing records – Developed NLP pipelines for automated medical document classification and information extraction using TensorFlow and spaCy – Delivered text mining and NLP solutions for extracting structured insights from unstructured clinical and financial documents, foundational work directly applicable to modern RAG and LLM pipelines – Applied deep learning models (RNN, LSTM) for sequential data pattern recognition in insurance fraud workflows – Implemented large-scale data pipelines on Azure Databricks (PySpark) and Azure Data Factory for multi-source integration across regulated environments – Built ETL/ELT pipelines on AWS (Glue, EMR, Redshift) for analytical processing of healthcare and energy datasets – Designed and industrialized batch and streaming data processing workflows for heterogeneous data sources – Ensured compliance with data governance and security requirements (HIPAA, KYC) in highly regulated contexts – Deployed ML models using MLflow and Docker, establishing model versioning and reproducibility standards across client projects – Implemented CI/CD pipelines for automated model deployment and monitoring – Acted as ML technical referent: validated model architectures, conducted code reviews, and advised on performance optimization – Defined technical standards, best practices, and industrialization patterns for client projects – Mentored and upskilled a team of 6 data and ML engineers, introducing best practices in model deployment, pipeline design, and data quality – Technically led client projects: task coordination, deliverables tracking, commitment management – Collaborated closely with client business teams to ensure solutions met operational needs