Naoufal Acharki

Led the end-to-end development of the Senzai recommendation engine, from concept to production, optimizing marketing campaigns with efficient customer acquisition, engagement and retargeting. – Causal Modeling: Developped a Causal modeling pipeline using Causal Random Forest and Meta-Learners with XGBoost for marketing uplift modelling. – Experimentation: prepared pre‑, during and post‑experimentation analysis with determining optimal test group sizes and minimum detectable effects. – LLMs/NLP: Applied Gemini and GCP Vertex.AI to analyze product descriptions and images (Embeddings) and create grouping for efficient personalized recommendations. – Hands-On: Wrote SQL scripts and contributed to senzai ETL pipeline by conducting in-depth exploratory analysis and unifying fields across source tables. – Led MLOps initiatives by ensuring best coding practices (including testing, documentation), architecting, deploying, integrating and maintaining workflows utilizing GitLab CI/CD, Docker, Apache Airflow and MLflow. – Integrated Slack notifications app with Gitlab to send Projects' updates including Pull Requests, Merges and Releases. – Led cross-functional problem-solving, aligning backend with frontend/DevOps requirements.

– Data production: Sourcing, querying, cleaning and analyzing large datasets, up to 50 million rows (e.g. buildings, roof slopes), from a datalake hosted in Google Cloud Plateform (GCP). – Statisticall modelling: Applying mathematical models and quasi‑experimental design approaches (e.g PCA with LHS) to improve the performances of existing algorithms – Decreased the computational time of panels packing algorithm by 3 times. – MLOps: Writing python/SQL recipes for production and deployment – Industrializing methods and models on modules/packages to ensure easy accessibility and maintainability of the database.