jaskaran5109/ai-proctored-interview

Purpose and Entry Points

This is a full-stack AI-proctored interview platform for candidate interviews, recruiter/session management, coding assessment, proctoring signals, and AI-assisted evaluation. The backend starts through backend/src/server.js, assembles Express application concerns in backend/src/app.js, and exposes APIs from backend/src/routes/index.js. The React frontend boots at frontend/src/index.tsx and selects application views in frontend/src/App.tsx.

Backend HTTP and Security Boundary

Express routes dispatch to domain controllers for authentication, interview/session operations, AI, and code execution: backend/src/controllers/auth.controller.js, backend/src/controllers/session.controller.js, and backend/src/controllers/code.controller.js. Authentication middleware protects applicable endpoints using JWT utilities, while rate limiting and centralized error translation form the request boundary: backend/src/middleware/auth.middleware.js, backend/src/middleware/rateLimit.middleware.js, backend/src/middleware/error.middleware.js.

Persistence and Domain Model

Database initialization is isolated in backend/src/config/db.js; model registration/relationships are centralized in backend/src/models/index.js. Controllers should use this model layer rather than directly coupling frontend workflows to storage. Seed support in backend/src/utils/seed.js indicates environment/bootstrap data may be required for usable interview flows.

Interview Orchestration

The interview lifecycle is coordinated as a service rather than embedded solely in controllers, with backend/src/services/interview-flow.service.js representing the primary workflow boundary. Session endpoints/controllers expose this lifecycle to clients through backend/src/controllers/session.controller.js. Live updates are likely supported through the socket utility, backend/src/utils/socket.js, separating real-time event wiring from REST request handling.

AI and Evaluation Integration

AI responsibilities are layered: backend/src/controllers/ai.controller.js accepts API-level requests, backend/src/services/ai.service.js implements application AI behavior, and backend/src/ai/ollama.client.js integrates with an Ollama runtime/service. This isolates prompts/model invocation from interview orchestration and enables AI assistance or evaluation without exposing the model provider directly to the frontend.

Code Assessment Execution

Code submission validation and execution are distinct from general interview orchestration: backend/src/validators/code.validator.js, backend/src/controllers/code.controller.js, and backend/src/services/code-execution.service.js. This is a material trust boundary because candidate-supplied code requires constrained execution infrastructure, even though the specific sandbox/runtime implementation is not represented in the tree.

Frontend Application Flows

The frontend uses page-level routes/views for authentication, dashboards, interview rooms, session details, and results: frontend/src/pages/AuthPage.tsx, frontend/src/pages/InterviewPage.tsx, and frontend/src/pages/ResultPage.tsx. API access is centralized through service modules such as frontend/src/services/http.ts, frontend/src/services/interview.ts, and frontend/src/services/sessions.ts, keeping UI components from owning transport details.

Client State and Interview Experience

Authentication bootstrap and persistent auth state are handled by frontend/src/hooks/useAuthBootstrap.ts and frontend/src/store/authStore.ts. Interview-specific monitoring and state are separated into frontend/src/hooks/useInterviewMonitor.ts and frontend/src/stores/interviewStore.js. The candidate/recruiter experience is composed from focused modules: live coding in frontend/src/components/interview/LiveCodingWorkspace.tsx, AI interaction in frontend/src/components/interview/AssistantPanel.tsx, and interview/session creation in frontend/src/components/dashboard/CreateInterviewWizard.tsx.