mohamed aloui

Designs and implements LLM evaluation tasks to stress-test multi-step reasoning, tool use, and edge-case handling, surfacing hallucinations and other failure modes. Builds reproducible datasets, grading rubrics, and scoring frameworks for cross-model comparisons and governance decisions.

Built and deployed LLM-powered chatbots and AI agents for multiple B2B clients, automating customer-service tasks and reducing manual handling time. Shipped conversational systems using FastAPI and improved production configurations using A/B tests on prompts, retrieval, and model variants.

Curated and annotated a 5,000+ image dataset for industrial quality-control, balancing normal and defective samples to prevent class imbalance. Fine-tuned a YOLOv12 model for manufacturing anomaly detection and built an OpenCV preprocessing and augmentation pipeline to evaluate performance across defect classes.

Led technical feasibility assessments for industrial projects using quantitative analysis and predictive modeling of complex requirements. Designed automated analysis platforms with VBA and coordinated quality control and continuous improvement efforts to increase manufacturing efficiency.

Delivered 10+ end-to-end AI projects focused on generative AI/LLMs, computer vision, MLOps, NLP, and data engineering.

Completed a Master of Engineering in Mechanical Engineering with a quantitative foundation in advanced statistics, linear algebra, numerical methods, optimization, and mathematical modeling.

Completed preparatory classes for engineering schools in Monastir, Tunisia.