What the Hill
Your face gives you away.
What the Hill is an interactive, hardware-powered game that challenges you to predict how composed you look under pressure, then puts that prediction to the test.
Ever wondered if you look as calm as you think you do during a stressful interview question? We built What the Hill to make that gap visible in real time.
Players use a physical rotary dial to predict how unreadable their facial expressions will be, then answer a randomly generated interview question on camera. Using Presage's facial expression analysis, the game measures the confidence of the neutral expression classification throughout the six-second round. An ElevenLabs AI narrator delivers a brutally honest verdict, comparing the player's prediction against the live reading.
The result is a mix of self-awareness, friendly competition, and a little embarrassment.
How It Works
- Predict: Turn the physical dial to choose a composure score from 0 to 100.
- Perform: Answer a surprise interview question while a webcam captures your facial expressions.
- Reveal: See your predicted composure, measured reading, prediction gap, and final score, accompanied by an AI-generated voice verdict.
Built With
What the Hill combines hardware and software into one interactive experience. An Arduino reads the rotary dial and button, while Python handles the game logic, scoring, and communication with the hardware. OpenCV processes the webcam feed, Presage provides live facial expression analysis, and ElevenLabs generates the spoken questions and verdicts. A React-based browser interface displays the live round, results, leaderboard, and calibration curve.
Why We Built It
We rarely get immediate feedback on how our facial expressions come across in high-pressure situations. We wanted to turn that invisible gap between perception and reality into something people could experience for themselves.
What the Hill isn't a stress detector, lie detector, or interview-training tool. It measures the model's confidence in classifying a neutral facial expression, not how nervous, honest, or capable someone actually is. The fun comes from seeing how closely your prediction matches what the system reads.
Think you've got a poker face? Let's find out.
Log in or sign up for Devpost to join the conversation.