Where I've worked
Part-time and internship roles — what I built, where I built it, and what each one taught me.
- Part-timePart-Time Machine Learning Engineer
Core Techs SolutionsSeptember 2025 — Present · Sousse, Tunisia- —Built an end-to-end insurance fraud detection pipeline by consolidating and cleaning fragmented claims data, engineering statistical risk features, and applying unsupervised anomaly detection (Isolation Forest, DBSCAN) to produce a composite fraud scoring system with investigation-ready visualizations surfacing the highest-risk cases.
- —Built an end-to-end document intelligence pipeline integrating PaddleOCR/LayoutLM for structured extraction and Qwen-VL for visual reasoning, with a custom C++ inference layer exposed to the JVM through Java FFM for native, high-performance execution.
- —Trained and benchmarked multiple object detection architectures (YOLO, RT-DETR, R-CNN) in PyTorch to assess used book condition, achieving a 25% accuracy improvement, optimized the best model to ONNX and accelerated inference with TensorRT, containerized the pipeline with Docker for reproducible, cost-efficient deployment. Exposed the model as a REST web service consumed by a Spring Boot backend, and developed the companion mobile frontend using Flutter.
- InternshipMachine Learning Engineering InternCore Techs SolutionsJune 2025 — July 2025 · Sousse, Tunisia
- —Designed and trained a baseline computer vision model in PyTorch to classify document condition (folds, tears,degradation), building the preprocessing and feature extraction pipelines that served as the foundation for subsequent production improvements.
Keep Moving