LLM Service & CV Generator Agent

[verdzy-kin]

Title: [BACKEND] LLM Service & CV Generator Agent

Integrate OpenAI/Claude API, implement CV Generator Agent with caching, cost tracking, and error handling.

[verdzy-kin]

Acceptance Criteria:

• LLMService class with OpenAI and Claude clients
• Environment variables for API keys (.env)
• Generate text completions function
• Generate embeddings function
• Redis caching for LLM responses (24-hour TTL)
• Cost tracking: log tokens used, API spend per request
• Rate limiting to prevent quota exhaustion
• CV Generator Agent:
  • Takes user_id, job_id (optional) as input
  • Fetches user profile from database
  • Fetches job description if provided
  • Generates tailored CV using LLM
  • Returns CV content, cover letter, key highlights
• Error handling for API failures (timeout, rate limit, invalid key)
• POST /documents/cv/generate endpoint returns task_id
• GET /documents/cv/generate/{task_id} returns status & result
• Unit tests with mock LLM
• Swagger docs updated
• Cost monitoring dashboard endpoint