Hamza El Belghiti

• Maintained and evolved a dashboard centralizing QLB customer quality data for the A330 and A350 programs • Accelerated the identification of recurring quality non-conformities by developing an LLM-based analysis system for QLBs from final cabin presentations • Reduced quality teams manual search time by designing a RAG pipeline that classifies QLBs by analyzing their description, automatically retrieving the applicable acceptance criteria (reference, page, conditions) and status (acceptable/non-conforming) • Designed an end-to-end ML pipeline using XGBoost to predict customer presentation dates at the A330/A350 FAL, achieving a MAE of 4.1 days and 87% of predictions within ±7 days, improving team planning and resource allocation Technologies: Skywise (Workbook, Pipeline Builder, Contour, Modeling, Workshop), Python, PySpark, NLP, Gemini, Prompt Engineering

- Migrated from the workflow orchestration tool Mage to Airflow - Assisted in the setup and deployment of the new Airflow environment - Rewrote the DAGs for the analytical pipelines - Migrated dashboards from Looker Studio to Metabase - Developed an AI Assistant Chatbot (Blent Copilot) to help students on the platform, answering questions about specific courses, exercises, and projects - Developed a process to automatically grade students' projects using a LLM - Created educational content and hands-on tutorials for Analytics Engineering with dbt - Developed hands-on project exercises to create ELT pipelines using tools like Airbyte, Airflow, dbt, and BigQuery - Created SQL assessment exercises to evaluate students' proficiency in SQL Tools: Airbyte, Airflow, dbt, BigQuery, Metabase, Langchain, HuggingFace (Mistral, Llama), OpenAI API, Pinecone

- Implemented automated ELT pipelines to feed the Customer Data Platform on the Data Warehouse from sources like Hubspot, Retool, Zoom, MongoDB etc. - Developed Reverse ETLs for marketing tools such as Hubspot and CustomerIO. - Constructed a sales dashboard to analyze customer data (clients and leads), and a marketing dashboard to analyze marketing strategies. - Consulting for DGAC : Automated the workflow of extracting complex pieces of information from PDFs of aerodromes in France using an LLM. - Wrote educational content about Analytics Engineering (dbt and Airflow) for the Data Engineering course. - Created SQL and Python Coding Challenges. Tools: Airbyte, Mage, BigQuery, dbt, Looker Studio, Open Metadata, CustomerIO API, Langchain, PDFplumber, OpenAI API, ChromaDB, Streamlit

- Deployed a 3D teeth segmenting deep learning model using a REST API on the AWS Cloud using Docker, AWS ECR, Lambda, and API Gateway. - Reduced deployment cost from $85 per month fixed to $1.3 per 1000 requests. - Created a user interface with 3D libraries (Vite, React.js, Three.js, and VTK.js) and deployed the web application with Vercel. 🖱️ Live Demo: https://teethseg-ai.vercel.app 💻Github Repository: https://github.com/Hamagistral/TeethSeg

- Collaborated with a team of 10 international people to develop an AI conversational chatbot that makes finding alternative public transport during strikes in the Île-de-France, Paris region easier. - Web scraping of the RATP site for real-time traffic status, and managed data uploads to an AWS S3 bucket with an ETL using the AWS Lambda function. - Created an algorithm that calculated the nearest available transportation station based on user preferences and location, reducing the impact of transportation disruptions in Île-de-France and offering personalized alternative routes and directions. - Implemented a version using the Navitia API, ensuring accurate and up-to-date transportation data retrieval by communicating directly with RATP's non-public API, enhancing the chatbot's reliability. 🖱️ Demo: https://youtu.be/MHPyRCB5QLk 💻Github Repository: https://github.com/Hamagistral/Omdena-france-strikes-transportation