Sara KANADI

• Led end-to-end delivery of a data-driven optimization model for OCP Group — Africa's largest industrial clients — modeling the problem, building the solution under real operational constraints, and presenting results to senior stakeholders. • Synthesized insights from 100+ sources into actionable research findings on AI adoption trends. • Contributed to 2 research papers on mono- and multi-production optimization models under industrial and sustainability constraints. • Collaborated with an interdisciplinary team of researchers and PhD students to validate model assumptions, structure findings, and communicate results to stakeholders. Teaching Assistant — Executive Master's in Business Analytics (MBAT) | Mar 2024 – Jun 2025 • Designed course materials, practical assignments, and grading rubrics for a cohort of 36 Executive Master's students, bridging advanced ML concepts with real-world business applications. • Mentored students through hands-on Python and Analytics projects, and managed classroom activities during practical sessions and addressed student questions. Collaborative Project — LLM-Powered Customer Sentiment& Churn Analysis • Collaborated with PhD researchers and data engineering intern to design and deliver a customer segmentation solution on Azure ML, applying K-Means and hierarchical clustering to 200K+ records with RFM feature engineering to surface 6 actionable behavioral segments. • Built a RAG pipeline with LangChain, FAISS, and HuggingFace models to augment customer profiles with historical context, then containerized the system with Docker and exposed it via a FastAPI service. • Presented segment profiles and ~20% campaign ROI uplift through an interactive Streamlit dashboard, translating analytical outputs into decision-ready recommendations for marketing stakeholders. Collaborative Project — Insurance Pricing ML System • Partnered with PhD researchers and a software engineering intern to scope and deliver a regression-based pricing model on 500K+ records (18% RMSE improvement), coordinating across modeling, deployment, and testing workstreams. • Containerizing the model with Docker, building a FastAPI wrapper while communicating progress and results to a mixed technical and non-technical audience. • Co-developed Streamlit dashboard to communicate risk distributions and business impact simulations, turning collaborative research output into shared, actionable business insight.

• Built an end-to-end NLP pipeline to construct knowledge graphs from unstructured text, using transformer-based models for Named Entity Recognition and Relation Extraction. • Engineered a data transformation pipeline that converted raw textual data into a structured, queryable knowledge base, reducing manual research time and supporting data-driven decision-making. • Evaluated and benchmarked multiple graph embedding algorithms; identified and documented the optimal production approach, improving downstream inference accuracy by 15%.

• Analyze customer baskets using Data Mining • Implement the KDD process on a transactional dataset • Extract association rules from clients’ purchases • Develop a web application to visualize the results