Zakaria DOUABI

Context: Strategic project for automating contractual processes within Orange Wholesale’s B2B division, aimed at improving contract reliability and reducing validation delays. Key Achievements: - Full project ownership of “Bot Contrat IA”: designed, developed, and deployed a semantic document analysis engine. - Developed the core application using LangChain, Python, AWS S3, ElasticSearch, and FastAPI. Implemented an ETL pipeline for contract ingestion, vectorization, and indexing. - Built a question-answering model based on LLMs (GPT / fine-tuned BERT) to automate contract-related responses. - Designed a performance monitoring dashboard with Power BI / Streamlit. - Led the functional scoping: business requirements gathering, user stories, technical documentation, and project committee reporting. - Conducted user training and onboarding on the use of generative AI for document workflow automation. Key Results: - 50% reduction in contract-related disputes - 60% decrease in administrative processing time - 92% user satisfaction rate - Estimated ROI of €300,000 per year - 80% adoption by operational teams within 3 months Tech stack: Python, LangChain, AWS S3, ElasticSearch, FastAPI, Power BI, MLflow, Docker, GitHub Actions, Agile/Scrum

- Design and implementation of a Feature-Aware Network (FAN) model for Channel State Information (CSI) prediction. - Performance optimization through reduction of OTA overhead and signaling latency. - Used PyTorch and TensorFlow for training multi-temporal recurrent neural networks. - Set up a simulated experimental environment using 5G Toolbox (MATLAB) and evaluated performance on an internal dataset. - Authored the technical report and presented results to the Nokia Research scientific committee. Results: - 40% reduction in OTA overhead - 30% decrease in average transmission latency - Prototype validated and integrated into the 5G/6G R&D research pipeline Tech stack: Python, PyTorch, TensorFlow, NumPy, MATLAB, Git, Jupyter, Linux, Snowflake

Context: Energy prediction project for the industrial sector (IoT / Smart Factory) aimed at cost optimization and carbon footprint reduction. Key Achievements: - Developed a predictive model for energy performance using real-time sensor data (time series). - Conducted exploratory data analysis, anomaly detection, and feature engineering (Python, Pandas, Scikit-learn). - Implemented machine learning algorithms (LightGBM, Random Forest, XGBoost, R). - Built a CI/CD pipeline for automated training and deployment (Docker, GitLab CI, MLflow). - Designed data visualization and model monitoring dashboards using Streamlit and Power BI. Results: - 47% improvement in forecast accuracy - 80% reduction in average on-site energy consumption - 25% decrease in overall operating costs within 3 months Tech stack: Python, LightGBM, Scikit-learn, R, Docker, MLflow, Streamlit, Power BI, GitLab CI/CD

-Design and development of a web application in JEE for the automatic integration of performance bonuses and variable elements into HR Access (HRIS). -Use of the Agile method for project management and delivery of high-quality solutions -Definition and documentation of the application's technical and functional specifications -Close collaboration with HR and IT teams for application implementation and integration -Test and validate the application to guarantee its reliability and performance

- Led and coordinated a team of IT technicians to deliver digital transformation projects across the organization. - Defined and tracked IT quality metrics to enhance system reliability and end-user satisfaction. - Designed and programmed smart electronic boards to automate the farmers’ compensation system. - Completely redesigned the company’s website (React.js & AngularJS) with enhanced performance, SEO, and cybersecurity. - Managed network security, preventive maintenance, and IT infrastructure administration across hybrid environments. Key Achievements: - Improved operational efficiency by 45% through optimized IT workflows. - Reduced network incidents by 50% via enhanced monitoring and proactive maintenance. - Achieved a 60% faster compensation processing time with the deployment of smart embedded systems. - Enhanced website performance by 40% and strengthened uptime reliability. - Increased user satisfaction to 95% across internal digital platforms.