Mahdi Miri

Introduction

I am an engineer passionate about bridging the gap between messy, real-world data and accurate, scalable intelligent systems. I specialize in connecting strong machine learning foundations with practical software engineering to solve complex problems. Currently, I am deeply focused on Large Language Models (LLMs) and Generative AI, exploring methods to fuse classic ML reliability with LLM reasoning to create models that are context-aware, reliable, and human-aligned. 🚀 Core Competencies & Focus Areas: Predictive Modeling & ML Engineering: Building high-performing end-to-end pipelines. Proven track record includes achieving R² = 0.96 in vehicle valuation using Stacking & XGBoost and modeling ALS progression via Ordinal Logistic Regression. Generative AI & NLP: Developing RAG-based applications, mastering prompt engineering, and experimenting with open-source LLMs. My goal is to integrate deep reasoning capabilities into practical AI solutions. Deep Learning & Optimization: Designing robust architectures (ANNs with embeddings, TabNet) and fine-tuning them with Optuna for high-dimensional forecasting tasks. Data Engineering: Crafting resilient pipelines capable of handling severe data imbalance (SMOTE/SMOTEENN), outlier control, and advanced feature engineering. From Models to Applications: Deploying models into interactive Streamlit apps for real-time insights and utilizing clustering techniques (K-Means, PCA) for behavioral analysis. 🛠 Tech Stack: Languages: Python , R ML & DL: TensorFlow, Keras, PyTorch, Scikit-learn, XGBoost, Statsmodels Techniques: RAG, Prompt Engineering, Feature Engineering, Dimensionality Reduction Visualization & Apps: Streamlit, Plotly, Seaborn I am always happy to connect and discuss AI architecture, ML systems, or new tech ideas. Feel free to explore my projects below! 🌐 Portfolio: https://mahdi-miri.github.io/my-resume/

Liens