2025–present / Independent project / Private project
Multimodal Safety Chatbot
A private multimodal AI project exploring safer, family-supervised communication practice for autistic individuals.
Private project — this page shares only a high-level description. No private repository or live link is exposed.
This is a privately developed project. The public description is intentionally limited to a high-level overview.
The system combines facial-expression recognition, RoBERTa-based sentiment analysis, and voice features such as pitch and stress to interpret conversational context. It also uses a locally hosted Ollama chatbot, periodic visual analysis, and a safety-scoring layer for multimodal signals, conversation content, and potential interaction risks.