Virtual Booth

Malaysian students frequently struggle to translate passive English comprehension into active speaking skills for high-stakes exams like MUET. This challenge is magnified in rural areas, where unstable connectivity renders standard cloud-based educational AI useless. Eloquence Lab (ELoLab) bridges this digital divide as an infrastructure-independent, "Sovereign AI" speech coach. Built natively on the Google tech stack, ELoLab utilizes TensorFlow Lite for on-device edge computing, enabling zero-latency, 100% offline grading of spoken responses. It evaluates structural coherence using examiner-approved frameworks without requiring an active internet connection, while utilizing Firebase for asynchronous educator syncing. By combining scenario-based learning with decentralized AI evaluation, ELoLab directly advances SDG 4 (Quality Education). It delivers a highly scalable, secure, and private learning ecosystem designed to confidently elevate marginalized students from MUET Band 3 to Band 5, regardless of their geographic limitations.