Sunny Chatbot Dashboard

Sunny is ASU's chatbot — the conversational interface that handled student inquiries across SMS, web, and mobile. The dashboard behind it is the operations surface where university staff monitored conversations, refined training data, and watched the metrics that justified the system's existence.

I built both halves: the conversational engine that students saw, and the React + AWS dashboard that staff used to keep it tuned. The conversational side meant integrating with NLP backends, handling fallback flows for ambiguous queries, and structuring the conversation tree so it could grow as more student-services categories were added. The dashboard meant making operational data legible to non-engineers — administrators who needed to see "where is Sunny failing" without learning a query language.

The outcome the institution cared about: $1M projected annual operational savings. The number lives in a presentation deck somewhere. What I cared about was the deflection rate — the percentage of student inquiries that resolved in the chatbot without escalation to human staff. That's the actual measure of whether the system worked.

[PLACEHOLDER — fuller detail on the NLP integration, the dashboard architecture, and the specific deflection rate gains coming.]