Developed and delivered a multimodal anomaly detection feature for autonomous driving in unstructured environments within the OpenHUTB/nn repository. The solution focused on enhancing safety and robustness by detecting unexpected scenarios through sensor fusion and deep learning techniques. Integrated into the Carla-based carla_auto_vision_navigator module, the feature processes diverse real-world signals to identify anomalies, supporting safer navigation decisions. The work demonstrated an end-to-end pipeline, from data ingestion to actionable detection output, and was implemented primarily in Python. This approach enables future scalability and the addition of new modalities, reflecting strong skills in computer vision, robotics, and autonomous systems engineering.