AI‐powered mock interview platform that turns your resume and a target job description into realistic, feedback‐rich interview practice sessions.

Live: https://fiterview.com/

• Overview
• Key Features
• Architecture
• Tech Stack
• Monorepo Layout
• Getting Started
  • Prerequisites
  • Environment Variables
  • Local Development
• Database
• Project Scripts
• Coding Standards
• Deployment
• Roadmap
• Contributing
• License

Fiterview helps candidates rehearse interviews end‐to‐end:

1. Upload or paste your resume.
2. Paste a job post (JD) you’re targeting.
3. Generate a tailored interview plan (question bank by topic/seniority, behavioral + technical).
4. Conduct a live mock with voice (STT/TTS) or text.
5. Receive structured feedback and an improvement report.

The app is built with a clean, layered architecture to keep domain logic independent from adapters (UI, APIs, vendors). The App Router organizes routes and server actions, while Prisma + PostgreSQL persist user data and reports.

• JD‐aware question generation – prompts are conditioned on both the job post and resume to remove generic "one‐size‐fits‐all" questions.
• Behavioral + technical coverage – STAR/ABC prompts, system design, React/Next.js/web‐fundamentals tracks.
• Live or async practice – answer via microphone (speech‐to‐text) or type; text‐to‐speech for interviewer voice.
• Rubric‐based scoring – clarity, correctness, seniority signal, and actionable improvements per answer.
• Session history – store attempts, scores, and feedback for longitudinal progress tracking.
• Privacy‐first design – user artifacts are scoped to the account; secrets stored in envs.

Note: Some features may be behind feature flags while the product evolves.

┌──────────────────────────────┐
│ App UI │ Next.js (App Router), React, Tailwind
│ (app/*, components, hooks) │
└──────────────┬───────────────┘
 │ Server Actions / API Routes (app/api/*)
┌──────────────▼───────────────┐
│ Application │ Use cases / services orchestrating flows
│ (backend/application/*) │
└──────────────┬───────────────┘
 │ Calls into domain + adapters
┌──────────────▼───────────────┐
│ Domain │ Pure business entities, value objects
│ (backend/domain/*) │
└──────────────┬───────────────┘
 │ Port interfaces
┌──────────────▼───────────────┐
│ Infrastructure │ Prisma repos, LLM/STT/TTS/email adapters
│ (backend/infrastructure/*) │
└──────────────┬───────────────┘
 │
 ┌───────▼─────────┐ ┌───────────────┐
 │ Prisma ORM │────▶│ PostgreSQL │
 │ (prisma/*) │ │ (DATABASE) │
 └─────────────────┘ └───────────────┘

1. User submits resume + JD → server action builds a PromptContext.
2. QuestionGenerator use case requests questions from the LLM adapter.
3. User answers via text or mic → EvaluationService composes scoring rubrics and LLM feedback.
4. Results stored via Prisma repositories.

• Web: Next.js (App Router), React 18, TypeScript, Tailwind CSS
• Auth: NextAuth.js (session/callbacks, providers)
• DB/ORM: PostgreSQL + Prisma
• AI: LLM(s) via provider SDK (configurable); planned multi‐model routing
• Speech: STT/TTS adapters (pluggable; e.g., Whisper‐like STT)
• State: React Query (server state) + Zustand (client state)
• Email/Notifs: Adapter‐based (e.g., Resend or SMTP)
• Tooling: ESLint, Prettier, commitlint, Husky

Exact vendors are abstracted via adapters in backend/infrastructure/* to keep the core portable.

.
├─ app/ # Next.js App Router pages, layouts, server actions, API routes
├─ backend/ # Domain / application / infrastructure (Clean Architecture)
├─ prisma/ # Prisma schema, migrations, seed
├─ public/ # Static assets
├─ hooks/ stores/ # Reusable React hooks, Zustand stores
├─ lib/ utils/ # Shared libs, helpers, validators
├─ constants/ types/ # App‐wide constants and TypeScript types
├─ next.config.ts # Next.js config
├─ middleware.ts # Auth / routing middleware
└─ package.json # Scripts and workspaces config (if any)

• Node.js: 18.x or 20.x LTS
• Yarn or pnpm (project uses Yarn by default)
• PostgreSQL: local or cloud instance

Create a .env file at the project root:

Some variables are optional depending on which adapters you enable.

# 1) Install deps
$ yarn install
# 2) Generate Prisma client
$ npx prisma generate
# 3) Provision DB & run migrations
$ npx prisma migrate dev --name init
# 4) (Optional) Seed local data
$ npx prisma db seed
# 5) Start dev server
$ yarn dev
# App runs on http://localhost:3000

If you change the Prisma schema, rerun prisma generate and apply a new migration.

• Prisma schema lives in prisma/schema.prisma.
• Use prisma/migrations/* for versioned changes.
• Seed script (prisma/seed.ts) can populate demo users and sample question banks.

Common commands:

# Inspect database in a browser
npx prisma studio
# Create a new migration from schema changes
npx prisma migrate dev --name <change>
# Push schema without creating a migration (non‐prod)
npx prisma db push

Package scripts (run with yarn <script>):

• dev – start Next.js in development
• build – typecheck and build production assets
• start – run the production server
• lint – run ESLint
• format – run Prettier
• typecheck – run tsc --noEmit
• prisma:* – convenience wrappers around Prisma CLI

See package.json for the authoritative list of scripts.

• ESLint with TypeScript rules; Prettier for formatting.
• Commitlint + Husky enforce conventional commits (e.g., feat:, fix:) on commit/push.
• Folder‐by‐layer structure; keep React components presentational where possible and push side‐effects into use cases/adapters.

• Recommended: Vercel for the Next.js frontend + serverless API routes.
• Ensure all required environment variables are set in the hosting provider.
• Use a managed PostgreSQL (e.g., Supabase, RDS, Neon) for production.
• Run yarn build && yarn start for a Node server deploy.