InterviewDojo: Voice Mock Interviews on Alexa+

• 

  InterviewDojo landing page: the Alexa+ voice simulator with resume presets, Streamable HTTP MCP and Amazon Bedrock scoring.

• 

  Interview setup: choose a target role, difficulty (up to 10 questions), then paste or load a resume and job description.

• 

  Start the spoken mock interview. Under the hood: resume and gap parsing, a voice interview via Web Speech API, and a weak-spot.

• 

  Live interview screen: Alexa+ asks a question aloud, the candidate answers by mic or typing, and the topic blueprint

• 

  Instant score card: each spoken answer is graded for correctness, depth and clarity, with feedback, a follow-up

• 

  End-of-session report: overall score, per-topic breakdown, strengths and weak spots, so candidates know exactly what to practise next.

• 

  InterviewDojo report showing topic scores, strengths, weak spots, a three-step practice plan and a button to start a new mock session.

Inspiration

Interview practice is generic. Question banks know nothing about the projects on your resume, and human mock interviewers are hard to find. My earlier AI Resume Analyzer only gave a score, so I wanted an interviewer that asks about my work, out loud.

What it does

InterviewDojo is a self-hosted MCP server that lets Alexa+ run a spoken mock interview. It asks questions from your resume and a job description, scores each answer on correctness, depth and clarity, and ends with a weak-spot report and practice plan.

How I built it

Node.js, TypeScript and Express with the MCP SDK (Streamable HTTP), Amazon Bedrock for question generation and scoring, MongoDB for sessions, and a React simulator with browser speech. The AI Resume Analyzer existed before the hackathon; the MCP server, scoring logic and voice simulator are new.

Challenges

Keeping questions tied to the candidate's real projects, getting consistent scores from the model, and learning MCP's Streamable HTTP transport.

What I learned

Designing MCP tools, getting structured JSON from Bedrock, and writing for voice instead of text.

What's next

Coding-round questions and progress tracking across sessions.

Built With

• alexa-plus
• amazon-bedrock
• amazon-web-services
• express.js
• git
• github
• javascript
• mcp-sdk
• model-context-protocol
• mongodb
• mongoose
• node.js
• react
• rest-api
• streamable-http
• tailwindcss
• typescript
• vite
• vitest
• web-speech-api
• zod