Amin Lasri

👨‍💻 About Me

I am an AI Developer and Data Scientist specializing in engineering high-performance retrieval pipelines and machine learning systems. Backed by a Master’s degree in Applied Econometrics and a Diploma in Artificial Intelligence from La Cité, I focus on transitioning complex quantitative models and data architectures into production-ready deployments.

My background bridges the gap between rigorous statistical inference and modern AI engineering, allowing me to build solutions that are not just technically advanced, but mathematically sound and directly aligned with real-world business logic.

Core Data Science & Applied Statistics Experience:

  • Statistical Communication: Leveraged 8 years of experience as an Economics and Statistics Educator to consistently translate complex econometric theories, causal inferences, and predictive models into clear, actionable insights for diverse, non-technical audiences.
  • Advanced Predictive Analytics: Engineered end-to-end geospatial valuation models using supervised ensemble methods (XGBoost, CatBoost) to forecast real estate trends. Executed rigorous hypothesis testing and data transformation across 80+ complex variables to achieve a highly accurate predictive R² score of 0.9230.
  • Quantitative AI Evaluation: Designed mathematical benchmarking pipelines to evaluate Large Language Model outputs. Utilized the RAGAS framework to score Context Precision (0.83) and engineered cross-encoder abstention mechanisms to mathematically prevent logical hallucinations in enterprise architectures.
  • High-Dimensional Data Optimization: Executed advanced vector operations and model tuning on the massive Amazon ESCI e-commerce dataset. Fine-tuned deep learning models (Matryoshka embeddings) to compress data dimensions by 91%, significantly optimizing inference costs while maintaining strict recall and nDCG performance metrics.

📬 Connect


My Projects

Housing Price Prediction

Engineered an end-to-end machine learning pipeline to predict real estate valuations by analyzing structural data and geospatial clustering.

Parking Vision System

Developed a deep learning vision pipeline designed to guide autonomous parking maneuvers and detect spatial obstacles.

Amazon ESCI Search Engine

Click here to change this text. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Ut elit tellus, luctus nec ullamcorper mattis, pulvinar dapibus leo.

Scroll to Top