Al Amir Kinann's Portfolio

Protein Biological Function Prediction: Utilized Microsoft Azure for model training and deployment, including setting up a CI/CD pipeline for automated AI/ML model deployment.

Loan Default Prediction: Developed a hybrid ensemble model using ML/DL (CatBoost + TabNet).

Designed Machine Learning and LLM (CamemBERT(a)-v2, Mistral medium 2, Gemini 2.5 Pro) models for automated classification and implemented RAG (Retrieval-Augmented Generation) architectures for NLP.

Accelerated data collection for safety hazard studies by 30% through enriching the EDF-Hydro database metadata.

Built a PyTorch multi-class semantic segmentation model (8 categories) for drone imagery using GRICAD.

Optimized performance by fine-tuning decoders of pre-trained models.

Created a Streamlit (Python) application for monitoring mechano-chemical synthesis.

Characterized the evolution of chemical properties through the analysis of acoustic energy levels (noise fluctuations).

Early Alzheimer's Detection via NLP: Developed 3 ML models and a RoBERTa model to analyze language and identify cognitive markers of dementia.

Implemented an energy forecasting algorithm for the GreEn-ER building (Grenoble) through predictive modeling of annual energy requirements.

Performed time-series processing and benchmarking of classical Machine Learning models.

Engineered a Python data processing program via reverse engineering of a proprietary system.

Streamlined diffraction analysis and image reconstruction based on captured energy flows, enabling the identification of physico-chemical properties of nanostructures.

Business management and administration (MBA).

School of Energy, Water, and Evironment Engineering (ENSE3-INP) & Phelma-INP. Specialized in signal, image processing, Communication systems and multimedia.

Machine Learning, Algorthm Theory and Deep Learning.

Higher National Diploma in Applied Physics & Instrumentation.