HireSignal

Inspiration

Interviewing is one of the most stressful and opaque parts of the job search. Candidates can spend weeks preparing, yet still have no clear sense of how they would actually perform in a real interview or how close they are to receiving an offer.

We were inspired by the lack of tools that combine realistic interview practice, job-specific context, and honest evaluation. Most platforms offer generic questions or vague feedback. HireSignal was built to answer a question every candidate asks but rarely gets a clear answer to:

“If I interviewed for this role today, how likely am I to get the offer — and why?”

What it does

HireSignal is an AI-powered interview preparation platform that:

Takes a job description as input
Generates role-relevant interview questions
Conducts interviews via text or voice
Uses AI to evaluate performance, providing:
- An overall interview score
- Likelihood of receiving an offer
- Clear strengths
- Specific areas for improvement

The goal is to make interview prep measurable, actionable, and personalized.

How we built it

HireSignal is built using a modern full-stack architecture:

Frontend: React for an interactive interview experience
Backend: Node.js and Express for API orchestration
AI:
- Google Gemini for question generation and interview evaluation
- ElevenLabs for text-to-speech (AI interviewer) and speech-to-text (user responses)

The system works as a pipeline:

Generate interview questions from the job description
Collect user responses (text or voice)
Transcribe voice input when needed
Combine responses into a structured transcript
Evaluate the transcript using an AI scoring rubric

Challenges we ran into

Prompt reliability: Getting consistent, structured evaluation output from the AI
Latency: Coordinating speech-to-text, text-to-speech, and evaluation without slowing the interview flow
Voice UX: Handling pauses, retries, and imperfect speech transcription
Scoring fairness: Balancing realism with encouragement so feedback is honest but motivating

Designing an evaluation that felt useful rather than discouraging was one of the hardest parts.

Accomplishments that we're proud of

Built an end-to-end AI interview flow in a short time
Successfully integrated real-time voice interviews
Designed a job-specific evaluation system, not generic feedback
Delivered clear, actionable strengths and weaknesses instead of vague advice
Created a product that feels genuinely useful for real job seekers

What we learned

Grounding AI evaluation in a job description dramatically improves relevance
Voice-based interfaces require careful UX handling beyond simple transcription
AI feedback is most valuable when it is structured and scoped
Small prompt changes can significantly affect output quality

We also learned how to design AI systems that feel supportive, not judgmental.