Secret Agent Interviewer

• 

  Be prepared. Be very prepared.

Inspiration

After the Valley, I worked for UC Berkeley and Stanford GSB, and saw first hand how "classroom learning" is inadequate for real-world conversations, including interviews and pitches.

What it does

Tests the candidate on subject matter, adjusting the difficulty level based on the quality of responses. Data are logged to Weights&Biases for analysis of individuals and group statistics.

How we built it

React+Tailwind, FastAPI, webkitSpeechRecognition, Google Gemini for AI-speech processing, Stytch for magic link. Weave for data recording and display.

Challenges we ran into

The usual - everything takes long than expected, and "we" were unable to deliver all the planned functionality and integrations. Very mysterious still unresolved bug with STYTCH.

Accomplishments that we're proud of

It does the basics of asking intelligent questions of varying difficulty, and with some polishing, will be a useful tool in the job candidate's arsenal.

What we learned

Working under pressure is the best!

What's next for Secret Agent Interviewer

Further polish. Adding more categories. Fixing and extending Stytch integration. Expanding Weave reporting.

Built With

• fastapi
• gemini
• python
• react
• stytch
• tailwind
• weave