The Team

We are a team of engineers and math students from Columbia, the University of Toronto, and Stanford.

Inspiration

I was domscrolling on Instagram one day - came across this Cluely thing, they claim no one can detect them. Their founder's slogan is

cheat on everything

We don't think is fair, a lot of us worked hard to get jobs.

Another problem we see is the lack of standards across interviewers and the additional workload on senior SWEs to review interviews.

So we make Candidly - Anti Cluely for Candidates, Databricks for Interviewers.

What it does

It does two things:

  1. It detects and shows suspicious gazing and typing patterns for the candidates.
  2. It is a video interviewing tool with template support, code questions, and an editor for interviewers and candidates.
  3. It records and looks for notable candidates and also provides insights for human resources managers on interviewers' performance and directions.

How we built it

P2P interview video call with WebRTC;

The WebGazer gaze-tracking algorithm is based on Brown Univeristy Research [1];

Gemini API on langchain for RAG and agentic summary;

ChromaDB implements a vector database for RAG for scalability;

Deepgram for text-to-speech transcription;

Next.js frontend on Vercel for a middleware serveless function for API calling, secrets management, and hosting;

Fast API for state management and user information retrieval;

,.... and, of course, about seven and a half cans of Red Bull

Accomplishments that we're proud of

We implemented a reliable app that achieves everything we proposed.

What's next for Candidly

Train a more advanced gaze algorithm to detect complex anomaly signals, improve the frontend experience for HR and hiring managers, enhance connectivity, and expand web-based anti-cheat features.

^ as per the goal of this project, this is 0% AI-written :)

References

[1] https://webgazer.cs.brown.edu/

Built With

  • chromadb
  • deepgram
  • fastapi
  • gemini-api
  • langchain
  • next.js
  • vercel
  • webgazer.js
  • webrtc
Share this project:

Updates