Houssam B.

Developing advanced deep learning systems across multiple modalities including vision, language, and

multimodal fusion with applications in generative AI and foundation models.

• Leading cross-functional deployment workflows, reducing model integration time from 8+ engineering hours to

a 20-minute automated process.

• Architected and maintained two pipelines: (1) an automated data labeling system that significantly reduced

resource requirements, and (2) a comprehensive KPI evaluation framework enabling systematic performance

monitoring and continuous model optimization across production environments.

• Conducting cutting-edge research in vision-language models, transformer architectures, and self-supervised

learning frameworks to solve complex real-world problems.

• Technical Stack and Expertise: PyTorch | C/C++ | Python | Docker | TensorRT | ONNX | Kubernetes |

Diffusion Models | LLMs | MLOps.

• Design cutting edge 2D/3D perception system capable of detecting and predicting the entire environment surrounding a car to navigate in complex environments with varying weather conditions.

• Tools, languages and keywords : PyTorch, Keras, TensorFlow, ROS, C/C++, Python, git, Linux, SLAM, 2D, 3D, machine learning, image processing, localization, detection, segmentation.

• Tools and languages and keywords : C/C++, Python, Qt, git, JIRA, UML, Linux, Windows, MBSE, SCADE, MVC

• Modular software architecture design (plugins)

• Incremental development, client meetings (Agile, Scrum)

• Application Maintenance/updates : Graphic editor for avionics data | C++ Qt