Inspiration Most people never get high-quality interview practice—friends are awkward or unskilled, and real experts are hard to access on short notice. We wanted a safe, always-on way to rehearse that still connects to human judgment when it matters.
What it does Friendly Interviewer runs voice-style AI mock interviews with feedback and analytics, and helps users book sessions with real industry interviewers. It’s one flow from solo practice to pro-level validation.
How we built it We used a React + TypeScript + Vite frontend with Tailwind/shadcn-style UI, Firebase for auth and data, and OpenAI (plus voice-related pieces) for conversation and analysis. We stitched routing, dashboards, booking, and results into a single cohesive product experience.
Challenges we ran into Balancing latency, cost, and reliability across speech, transcription, and LLM analysis while keeping the UX smooth. Designing clear trust and safety boundaries between “AI practice” and “paid human sessions” without confusing users.
Accomplishments that we're proud of We shipped an end-to-end demo that feels like a real product—not a slide—covering practice, results, and community-style booking. We’re especially proud of the clarity of the user journey from first session to actionable feedback.
What we learned Product wins when feedback is immediate and specific; generic scores don’t change behavior. Integrating multiple AI/voice services teaches you to design for failure modes, retries, and empty states early.
What's next for Friendly Interviewer Add smarter role- and company-specific interview packs, stronger interviewer verification and scheduling, and deeper progress tracking over time. Longer term, partner with career centers and bootcamps to turn practice into measurable hiring outcomes.
Built With
- eleven
- fastapi
- labs
- typescript
Log in or sign up for Devpost to join the conversation.