samir ait-abbou

🔍 As an AI Engineer and Data Scientist, I develop innovative AI solutions to address research needs while supporting the digital transformation of organizations. •Conducted audits and specialized diagnostics to assess AI and data processing needs. •Developed functional prototypes (PoC) to validate the feasibility of AI-driven solutions. •Delivered training and technical support to business and engineering teams on AI and data science solutions. •Processed large-scale structured and unstructured data using Big Data tools (Spark, Airflow). •Designed and implemented MLOps pipelines: automation, CI/CD, containerization, and model monitoring. •Developed and deployed an administrative AI agent based on LLMs, Retrieval-Augmented Generation (RAG), and prompt engineering. •Built production-ready AI service APIs with FastAPI and deployed them on AWS using Docker. •Contributed to cross-functional activities: technical documentation, presentation materials, and AI technology watch.

🔍 Benchmarking Coding LLMs for Code Generation, Documentation, and Debugging Tasks: - Evaluated and compared coding LLMs (CodeLlama3, GPT-3.5, Codex, GPT-4) for code generation. - Measured model performance using precision metrics (pass@K). - Developed a data lifecycle pipeline to streamline model benchmarking and analysis. 🚀 Pipeline Implementation: - Data Streaming with Kafka: Set up brokers, schema management, and centralized control. - Advanced Data Processing: Managed data flow and tracked LLM metrics (pass@K) with Spark. - Efficient Data Storage: Implemented Cassandra (NoSQL) for fast and reliable data management. - Model Evaluation: Analyzed and stored LLM evaluation metrics in BigQuery. - Workflow Automation: Automated processes with Airflow and managed metadata in PostgreSQL. - Service Deployment: Containerized services on GCP using Docker. 🔧 Pipeline Development and Integration: - Built and maintained data integration and transformation processes. - Integrated the pipeline into CI/CD workflows using Azure DevOps and GitLab CI. - Developed technical and functional documentation to ensure process clarity. - Created test plans and conducted unit testing for quality validation. 🌟 Impact: These initiatives improved the efficiency of code generation benchmarks and enabled robust data lifecycle management for AI model performance evaluation.

🔍 Developed a data pipeline for the exploitation of geospatial data to study total water storage from multiple sources in Morocco. 🚀 Pipeline Implementation: - Lambda Architecture: Set up for batch and real-time processing using Spark. - Real-Time Data Ingestion: Utilized Kafka to stream geospatial data from a dedicated platform. - Data Storage: Stored data in HDFS using the Parquet format for multidimensional data management. - Data Cleaning and API Development: Cleaned raw data and created APIs using Spark for data consultation via Jupyter. - Monitoring and Visualization: Designed a Grafana dashboard to monitor Spark cluster metrics (failing nodes, resource consumption, alert logs, and errors). 🌟 Impact: Enhanced the accuracy of water storage analysis through real-time data processing and efficient geospatial data management.

🔍 Developed a data pipeline to monitor and analyze power outages in the Eastern region of France. 🚀 Project Implementation: - Data Cleaning: Processed and cleaned raw data, handling missing values and anomalies. - Data Preprocessing: Applied Principal Component Analysis (PCA) for dimensionality reduction and data interpretation. - Exploratory Data Analysis (EDA): Conducted in-depth analysis of outage patterns and factors. - Data Visualization: Analyzed power outages by type, category, location, and nature in cities (Champagne-Ardenne, Picardie, Lorraine), departments (Meuse, Vosges), and across different time periods (years, semesters, months, days, hours, minutes, seconds). - Cause Analysis: Studied outage causes including climate factors, residential density, and natural phenomena. 🌟 Impact: Improved the understanding of outage patterns and provided actionable insights for infrastructure management and failure prediction.

🔍 Developed an AI pipeline for the classification of scientific texts using Natural Language Processing (NLP) techniques. 🚀 Project Implementation: - Data Preparation: Performed segmentation, tokenization, lemmatization, stemming, and stop words removal. Vectorized text data using TF-IDF, GloVe, Word2Vec, and BERT Word Embedding. - Modeling: Implemented various neural network architectures including Neural Networks (NN), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Encoder-Decoder models, LSTM/GRU, and transformers (BERT, GPT-2). Also developed machine learning models for classification. - Deployment: Deployed the solution using Streamlit for interactive and user-friendly model inference. 🌟 Impact: Enhanced the classification accuracy of scientific documents, making data categorization more efficient and reliable.

🔍 Developed foundational skills in machine learning with a focus on supervised and unsupervised learning techniques. 🚀 Key Concepts Covered: - Introduction to Machine Learning: Understanding core concepts and applications. - Supervised and Unsupervised Learning: Implemented algorithms for both paradigms, including classification, regression, clustering, and dimensionality reduction. - Data Exploration and Analysis (EDA): Applied statistical methods to explore data, identify patterns, and detect anomalies. 🌟 Impact: Built a solid foundation in machine learning techniques and data analysis, paving the way for advanced AI projects.

🔍 Payroll Management Application Development: - Designed and developed a payroll management application to handle employee management, absences, expenses, bonuses, and pay slips. - Implemented backend functionalities using Laravel. - Created an intuitive user interface with Bootstrap. - Managed data storage and retrieval with MySQL. - Focused on performance optimization and secure data handling. 🌟 Skills Gained: 💬 Communication: Enhanced interpersonal and team collaboration skills. 🛠️ Problem Solving: Applied analytical thinking to backend architecture. 📊 Data Management: Efficiently handled payroll data and secure processing. 💡 Technical Proficiency: Gained hands-on experience with Laravel, Bootstrap, and MySQL.