Sohaila Abdulsattar Mohammed

Taught applied AI and machine learning courses, covering topics such as supervised learning, neural networks, model evaluation, and data visualization with tools like TensorFlow, Keras, NumPy, and Pandas. Led live coding sessions that walked through end-to-end model development, building training pipelines, tuning performance, and exploring datasets with real-time visualizations. Designed and adapted technical content to ensure conceptual clarity without oversimplification, tailoring explanations for varied programming backgrounds.

Led the design and development of a real-time sign language translation system integrating computer vision and language models for American and Arabic Sign Language. Built a sign-to-text inference pipeline combining SMPL body pose data with CLIP-style video-text embedding techniques for improved temporal alignment and gesture grounding. Collaborated with Deaf and Hard-of-Hearing partners to iteratively test, evaluate, and improve translation accuracy and ensure usability in real-world interactions.

Developed a conversational AI agent for a robotic barista for NYUAD’s RoboCafe project, aimed at creating a fully automated cafe experience where a robot prepares coffee based on customer voice commands. Integrated Pipecat for data piping, MeetKai Functionary for the large language model, and Tacotron2 for text-to-speech, managing both the interface and backend development for a voice conversational AI system. Contributed to weekly lab meetings by presenting detailed project updates and communicating challenges and solutions.