~/project/honkhonkhire.md
HonkHonkHire
Interview practice that watches back
- date
- September 2023
- type
- Hackathon
Donald Lee, with Vaibhav Sharma, Michelle Kim, Maureen Luo
## overview
It's job search season (again). The behavioural interview is the most nerve-wracking part of applying, and there's no good way to practice for it. Most tools can't emulate a real interview because they don't know anything about you, and real interviewers rarely give the kind of feedback that tells you what to fix.
HonkHonkHire is an AI interview coach. It reads your experience from your LinkedIn, resume, or portfolio, generates questions tailored to your story, then evaluates your facial expressions and transcribes your answers to hand back feedback you can actually act on.
## demo
## challenges
Challenges
The project was ambitious for the time we had: a lot of unfamiliar technology to learn and apply in a weekend. We had never worked together as a team before, so part of the weekend was just figuring out each other's strengths, weaknesses, and communication styles.
## what-we-re-proud-of
What we're proud of
We shipped most of the MVP, and we built something all four of us would genuinely use for our own job searches. Picking up Cohere's API and OCR Space mid-hackathon and actually getting them into the app was the rewarding part.
## what-we-learned
What we learned
Everyone stretched. Our designer wrote her first lines of HTML and CSS. Our front-end developer stuck to vanilla JS to drill the fundamentals instead of reaching for a framework. One developer learned new APIs from their docs and moved data between front and back end with Node.js. And I took on facial and emotional expression detection for the first time.
## what-s-next
What's next
We'd love to add richer metrics: talking pace, response length, tone of voice. With those, users could track their progress over a whole job search instead of a single mock interview.
## next-project
TravoAI
Trip planning without the tabs
read the write-up