Hamidoullah Hamidou Garba

During my internship within the IT department of the Rockefeller School in Lyon, I designed and developed a predictive dashboard for student performance to support instructors’ decision-making and improve academic monitoring. The project was based on the exploitation and integration of five datasets from three distinct sources (Aurion, Forms, MyRock), covering heterogeneous data such as grades by course unit and period, socio-economic profiles, entry-level assessments, internship competency tracking, and pedagogical interview reports. I implemented the entire data preprocessing pipeline, including data cleaning, handling missing values, normalization, descriptive and exploratory statistical analyses, as well as concrete recommendations to improve data quality and reliability at the source. Based on this, I developed ML Hub, a Flask-based platform integrating two machine learning applications: StatutPredict, which predicts a student’s academic status (passed, failed, dropout, on hold) based on socio-demographic data, and EduPredict, which evaluates academic level by field (low / middle / high) based on grades, with support for multi-track pipelines and configurable thresholds (up to 57 course units for nursing studies). Each application covers the full lifecycle, including model training (SMOTE, GridSearchCV, feature selection), model management and comparison, and deployment for prediction. Finally, I developed RAG Studio, an agentic RAG application that enables teams to query their document bases (PDF, DOCX, XLSX, etc.) via an AI chat interface, with support for multiple knowledge bases, web search, and conversation history. This internship allowed me to carry out end-to-end data and AI projects, from data collection and preparation to the deployment of user-facing interfaces, within a concrete and demanding educational environment.

AI & Data Engineer specializing in LegalTech, with expertise in integrating Large Language Models (LLMs), intelligent data extraction, and the automation of complex legal processes. • Designed and integrated a legal recommendation system based on audit results, enabling the automated generation of professional legal memoranda. • Developed advanced data extraction solutions from heterogeneous legal documents (shareholder registers, individual records, shareholder accounts), leveraging prompt engineering techniques and strictly structured JSON outputs to ensure data reliability and quality. • Implemented complex business rules related to legal compliance (consistency checks across registers, share movement calculations, identification of active shareholders, and share capital analysis). • Built robust ETL pipelines for ingesting, processing, and structuring large-scale data from multiple sources (INSEE, Infogreffe, Pappers), with full automation of the data lifecycle. • Developed Python scripts for data collection, cleaning, enrichment, and consolidation (pandas, requests), including error handling, API rate limiting, and performance optimization. • Designed and optimized both relational (PostgreSQL, SQLite) and NoSQL (MongoDB) databases, implementing batch processing to improve performance on large datasets. • Developed an automated legal report generation system (Word) with advanced formatting, integrating risk analysis and structured summary tables. • Integrated LLMs using LangChain within a multi-model architecture (AWS Bedrock, Google Gemini), implementing pipelines for structured text generation and response parsing. • Contributed to the development of an intelligent platform for document processing and legal decision support.

Stage Description : Developed an intelligent document management web application (ArchiveAI) at SmartIT, integrating AI capabilities to automate the full document lifecycle — from upload and classification to indexing, search, and archiving. Key achievements include: • Integrated Google Gemini API to build an automatic document classification system with two modes (strict & adaptive), including confidence scoring • Built an AI-powered indexation pipeline that automatically generates tags and summaries from document content (PDF, DOCX, XLSX, and more) • Designed and implemented an intelligent search engine leveraging weighted field scoring, Levenshtein similarity, and keyword normalization for fuzzy matching • Deployed a RAG (Retrieval-Augmented Generation) pipeline using LangChain and ChromaDB to enable natural language Q&A over archived documents • Developed a full REST API with Flask, covering authentication, role-based access control, and file integrity verification (SHA-256 hashing) • Built a real-time analytics dashboard with document statistics, indexation trends, and user activity tracking

As part of my internship at MPTC, I contributed to the creation of professional training video content in both French and English, focusing on aviation safety and security standards. Key Contributions Video Production & Editing • Designed and edited professional training videos using CapCut and Canva. • Ensured high-quality visuals and animations to enhance learning engagement. Aviation Safety & Security Training • Developed content based on ICAO Annexes, SMS (Safety Management System), SeMS (Security Management System), and Dirty Dozen (Human Factors in Aviation Maintenance). • Assisted in structuring educational materials aligned with international aviation standards. Collaboration & Teamwork • Worked closely with subject matter experts to ensure content accuracy. • Participated in team discussions and PSAT (Professional Skills & Assessment Training) sessions to improve training effectiveness. Tools & Skills • Video Editing: CapCut, Canva • Aviation Standards: ICAO Annexes, SMS, SeMS, Dirty Dozen • Soft Skills: Teamwork, Communication, Training Development Skills Developed • Instructional design • Multimedia content creation • Aviation safety & security compliance • Teamwork in a professional training environment

As part of my final-year internship in Software Engineering and Information Systems at the Faculty of Sciences of Monastir, I contributed to the development and enhancement of UnivAd, an ERP platform designed for university administrative management. • Project Objective Optimize and expand UnivAd to facilitate the management of study plans, academic years, faculty members, and students, reducing repetitive tasks and improving user experience. • Technical Contributions • Optimization of Study Plan Management Automated the creation and archiving process for study plans. Added a feature to duplicate and update study plan versions efficiently. • Development of the Student Management Module Designed and integrated a comprehensive student management module. Implemented student assignment to academic groups (lectures, tutorials, lab sessions, etc.). Developed an import/export system for student data using Excel files. • Enhancement of Academic Year Management Enabled the duplication of academic structures from one year to another. Simplified faculty and workload management processes. • Integration of Study Plan Synchronization with the Official Salima Platform Extracted study plan data through web scraping and automatically structured it within the database. • Technologies and Tools Used Backend: ADO.NET, ASP.NET Frontend: ASP.NET Database: SQL Server Others: REST APIs, web scraping, Excel (import/export), Git • Skills Developed Full-stack development in .NET, database management and optimization with SQL Server, API integration, business process automation, agile project management.