Salsabil GUIZANI

Architected and implemented RAG pipelines using LLMs and Azure OpenAI, combining semantic and hybrid retrieval, optimized chunking strategies, and re-ranking for improved answer accuracy. Developed production-grade PromptFlow pipelines integrating vector search, structured data sources, and enterprise documents for real-world business use cases. Built scalable invoice benchmarking pipelines with Azure Document Intelligence and Azure Blob Storage; implemented canonicalization and normalization of OCR outputs (VAT breakdown, business identifiers, language detection); designed automated evaluation with fuzzy matching and performance metrics.

Built YOLOv5-based object detection models for quality inspection in industrial environments. Developed full data pipelines for image processing and defect segmentation. Created a real-time application using Python, Streamlit, and FastAPI to visualize and process LabVIEW data.

Performed statistical analysis on operational data (call volumes, wait times, agent performance). Wrote optimized SQL queries and created automated dashboards with PostgreSQL. Improved staffing strategies through data insights and performance metrics.