Arya Araban

Designed complex math problems from images to reveal visual-understanding gaps in SOTA multimodal models. Analyzed reasoning patterns in LLMs by designing targeted failure cases across logic and coding domains.

Pioneered RangeSAM, the first ML model enabling promptable segmentation of sparse 3D LiDAR point clouds. Integrated range projection with transformer architectures to enhance autonomous driving scene understanding.

Served as Lead TA for multiple undergraduate courses and a graduate-level Computer Vision course. Delivered weekly technical lectures that covered both theoretical concepts and implementation methods.

Led vision efforts in cross-functional robotics projects, using computer vision for object and trajectory tracking. Achieved precise 3D object representation through stereo camera calibration and reconstruction techniques.

Engineered scalable query graph generators for a complex knowledge base, reducing query times by 30%. Analyzed and processed large datasets of news articles to identify patterns and extract key trends.

Developed and deployed Laravel applications with MySQL databases to ensure robust performance and security. Streamlined development with CI/CD pipelines and Agile methods to accelerate solution delivery timelines.

Pursued a Master of Science in Computer Science. Focused on advanced topics and research within the field.

Completed a Bachelor of Engineering in Computer Engineering. Gained foundational knowledge and practical skills in computer engineering principles.