feat: AI-Powered Interview Platform Foundation

[theashishmaurya]

🚀 Massive Update: AI Interview Platform Implementation

This PR implements multiple features from the detailed issue breakdown.

---

Phase 1: AI Code Helper ✅

#43 - OpenAI/Anthropic API Integration

import { getAIClient } from '@/lib/ai';

const ai = getAIClient();

// Streaming completion
for await (const chunk of ai.stream(messages)) {
  console.log(chunk.content);
}

// Single completion
const response = await ai.complete(messages);

Features:

• ✅ OpenAI GPT-4 support
• ✅ Anthropic Claude support
• ✅ Streaming responses (SSE)
• ✅ Rate limiting per provider
• ✅ Error handling with retries

#44 - Code Context Extraction

import { extractContext, buildPromptContext } from '@/lib/ai';

const context = extractContext(code, cursorLine, cursorColumn, 'typescript');
// { currentFunction, linesBefore, imports, ... }

const prompt = buildPromptContext(context);

Features:

• ✅ Extract current function/class at cursor
• ✅ Multi-language support (JS, TS, Python)
• ✅ Import extraction
• ✅ Surrounding context (20 lines before/after)

#45 - AI Hint System (Progressive 3-Level)

<HintPanel
  challengeId="modal"
  challengeDescription="Build an accessible modal..."
  currentCode={code}
  onHintUsed={(level, hint) => console.log(level, hint)}
/>

Features:

• ✅ Level 1: Direction hint (no code)
• ✅ Level 2: Implementation hint (1-2 lines)
• ✅ Level 3: Detailed hint (algorithm + edge cases)
• ✅ Per-user rate limiting (50/hour)
• ✅ Hint history stored in DB

#46 - Code Explanation Feature

<ExplanationPanel
  code={selectedCode}
  language="typescript"
  onExplanation={handleExplanation}
/>

Features:

• ✅ Streaming explanations
• ✅ Follow-up questions
• ✅ Copy to clipboard
• ✅ Markdown formatting

---

Phase 2: Interview Sessions ✅

#47 - Database Schema

Tables Created:

Table	Purpose
organizations	Team management
interviews	Interview sessions
interview_participants	Invites & participants
interview_sessions	Per-challenge tracking
invite_tokens	Secure invite links
hint_usage	Hint tracking
anti_cheat_events	Integrity logging
organization_members	Team memberships

Features:

• ✅ Full RLS policies
• ✅ Automatic timestamps
• ✅ Indexes for performance
• ✅ Secure invite token generation

---

API Routes

Route	Method	Description
/api/ai/completions	POST	Chat completions (streaming)
/api/ai/hints	POST	Generate progressive hint
/api/ai/hints	GET	Get hint history
/api/ai/explain	POST	Explain code (streaming)

---

File Structure

lib/ai/
├── client.ts          # AI client (OpenAI + Anthropic)
├── config.ts          # Configuration
├── prompts.ts         # System prompts
├── context-extractor.ts # Code context extraction
└── index.ts           # Exports

app/api/ai/
├── completions/route.ts  # Chat completions
├── hints/route.ts        # Hint generation
└── explain/route.ts      # Code explanation

components/editor/ai/
├── hint-panel.tsx        # Hint UI component
├── explanation-panel.tsx # Explanation UI
└── index.ts              # Exports

supabase/migrations/
└── 003_interview_sessions.sql  # Full schema

---

Environment Variables Required

# AI
OPENAI_API_KEY=sk-...
OPENAI_MODEL=gpt-4-turbo-preview
ANTHROPIC_API_KEY=sk-ant-...  # Optional
AI_MAX_TOKENS=4096
AI_TEMPERATURE=0.7
AI_RATE_LIMIT=50

# Supabase
NEXT_PUBLIC_SUPABASE_URL=...
NEXT_PUBLIC_SUPABASE_ANON_KEY=...
SUPABASE_SERVICE_ROLE_KEY=...

---

Usage Examples

Generate Hint

const response = await fetch('/api/ai/hints', {
  method: 'POST',
  body: JSON.stringify({
    challengeId: 'modal',
    challenge: 'Build an accessible modal...',
    currentCode: code,
    level: 2,
    previousHints: ['Level 1 hint...']
  })
});

const { hint, nextLevelAvailable } = await response.json();

Explain Code

const response = await fetch('/api/ai/explain', {
  method: 'POST',
  body: JSON.stringify({
    code: selectedCode,
    language: 'typescript',
    detailLevel: 'detailed'
  })
});

// Stream the response
const reader = response.body.getReader();
// ... handle SSE stream

---

Remaining Work

See issues for remaining tasks:

• Phase 2.2: Interview Creation Flow #48 - Interview Creation Flow UI
• Phase 2.3: Invite System & Email #49 - Invite System & Email
• Phase 3.1: Anti-Cheat UI & Warnings #50 - Anti-Cheat UI & Warnings
• 🎯 Phase 4: Live Video & Audio #35-41 - Phases 4-10

---

Acceptance Criteria

• AI client works with OpenAI
• AI client works with Anthropic (configurable)
• Streaming responses work
• Rate limiting enforced
• Hints are progressive (3 levels)
• Hint history stored in DB
• Code explanation streams to UI
• Database schema created
• RLS policies working

---

Relates to: #43, #44, #45, #46, #47
Master Tracker: #42

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
acecodinglab	Error		Feb 23, 2026 7:31pm