Job Analyzer

Job Analyzer is a backend tool that helps job seekers evaluate whether a job offer is worth applying to, using LLM-powered analysis of both the offer text and the company's reputation.

Problem

Job hunting is time-consuming and often demoralizing. Many offers look good on the surface but hide red flags: toxic culture, unrealistic expectations, or poor career growth. Manual research (Glassdoor, LinkedIn, news) is slow and inconsistent.

Solution

Job Analyzer uses an LLM to:

• Analyze the offer text against personal criteria (remote, stack, role, salary range).
• Research the company's reputation (culture, reviews, red flags).
• Return a clear recommendation with reasons, red flags, and positive signals.

Architecture

1. User submits an offer text and/or company name via REST API.
2. The service uses an LLM (LangChain) to extract key signals.
3. A company reputation module searches the web for reviews and signals.
4. The result is combined into a final recommendation.

Technologies

• Python
• FastAPI
• LangChain
• OpenAI / LLM APIs
• Pydantic
• Docker
• (pendiente: frontend ligero)

API Endpoints

Analyze a job offer

POST /api/v1/analyze

Request:

{
  "offer_text": "paste the full job offer text here"
}

Response:

{
  "should_apply": false,
  "reasons": ["company type does not match preferred criteria"],
  "summary": "Remote Python role at a global services company",
  "red_flags": ["top global clients", "project success rate"],
  "salary_info": null
}

Analyze a company's reputation

POST /api/v1/company

Request:

{
  "company_name": "My business imaginary"
}

Response:

{
  "company_name": "My business imaginary",
  "reputation_score": "negative",
  "summary": "Mixed reputation with concerns about culture and leadership",
  "red_flags": ["poor culture ratings", "low career opportunities"],
  "positive_signals": ["interesting projects", "flexible schedule"],
  "sources_consulted": ["https://glassdoor.com/..."]
}

Full analysis (offer + company)

POST /api/v1/analyze/full

Request:

{
  "offer_text": "paste the full job offer text here",
  "company_name": "Company Name"
}

Response:

{
  "offer_analysis": { ... },
  "company_analysis": { ... },
  "final_recommendation": "Not recommended to apply"
}

How to run it

docker-compose up --build

Or in local:

pip install -r requirements.txt
uvicorn main:app --reload

What I learned

• Designing a REST API with FastAPI for LLM-powered analysis.
• Using LangChain to structure prompts and parse responses.
• Integrating external data (company reviews) with LLM reasoning.
• Building a tool that solves a real problem I faced during my own job search.

Future work

• Add authentication and user profiles with personal criteria.
• Add a lightweight frontend (React) for easier use.
• Cache company reputation results to avoid repeated web searches.
• Add support for multiple LLM providers (OpenAI, Anthropic, local models).