04 / AI / PRODUCTIVITY

Interview Hint

A voice-driven interview preparation assistant that turns short technical hints into concise, structured explanations.

OVERVIEW

Built around the real problem.

Interview Hint captures a short voice hint such as 'vlan', checks local knowledge first, and returns a one-line definition, a 30-second answer, follow-up points, and a practical example. When local knowledge is insufficient, it can use OpenAI as a fallback.

KEY FEATURES

Profile input for role and technology context

Voice hint capture and transcription

Local fallback knowledge base before AI lookup

Structured answer validation

Recent hint cache

Windows self-test for sample hints

Android offline knowledge editing

Android headset, media, and volume triggers

Android profile relevance filtering

ARCHITECTURE

Windows Streamlit app: voice capture -> transcription -> cache and local JSON knowledge -> OpenAI fallback -> structured answer validation -> displayed response. Android Kotlin / Jetpack Compose app adds SpeechRecognizer, MediaSession, accessibility services, OkHttp, Moshi, and JSON storage.

RESULTS

Windows Streamlit MVP implemented

Android Kotlin / Jetpack Compose MVP implemented

Code inspection confirms Windows self-test and answer validation logic

Code inspection confirms Android speech, API, MediaSession, receiver, and accessibility paths

Android hardware behavior still requires testing on a real phone and Bluetooth headset

Android build was not verified from the previous check

Any hardcoded OpenAI API key must be revoked and removed before publication

MEDIA