Job Fit Assessment

Automated candidate-job fit evaluation powered by CrewAI agents. Provide a job posting URL and a resume PDF, and three AI agents will extract requirements, analyze the candidate, and produce a structured fit report.

How It Works

The system runs a CrewAI Flow with three sequential agents:

1. Skill Extraction Specialist — Scrapes the job posting URL using Firecrawl and extracts the job title, company name, and required skills.
2. Resume Analyzer — Extracts the full text from the candidate's PDF resume using PyMuPDF and scores the candidate against the extracted requirements, identifying strengths and gaps.
3. Report Writer — Compiles findings into a structured markdown report with fitness score, strengths, missing skills, and an overall assessment.

Progress updates are delivered in real-time via webhooks and Server-Sent Events (SSE) — the flow's event listener POSTs step updates to the frontend, which pushes them to the browser as each agent starts and finishes.

Project Structure

template_job_fit_assessment/
├── src/template_job_fit_assessment/
│   ├── main.py                # CrewAI flow with 3 agents
│   └── events/
│       └── listener.py        # WebhookEventListener (posts step updates)
├── frontend/
│   ├── app.py                 # Flask server (webhook receiver + SSE)
│   ├── templates/index.html   # Web UI
│   ├── static/style.css       # Styling
│   ├── requirements.txt       # Flask dependencies
│   ├── Procfile               # Heroku deployment
│   └── .env                   # API credentials (not committed)
├── pyproject.toml             # CrewAI project config
└── .env                       # API keys (not committed)

Prerequisites

• Python >= 3.11
• uv package manager
• API keys:
  • OPENAI_API_KEY — for the LLM agents
  • FIRECRAWL_API_KEY — for scraping job postings
• A Heroku account for the frontend
• A CrewAI AMP deployment for the backend

Deployment

1. CrewAI AMP (backend)

crewai deploy create

Set these environment variables in AMP:

• OPENAI_API_KEY
• FIRECRAWL_API_KEY
• WEBHOOK_URL — your Heroku app's webhook URL (see below)

2. Heroku (frontend)

# From the project root (template_job_fit_assessment/)
git subtree push --prefix frontend heroku main

Set these config vars on Heroku:

heroku config:set CREWAI_API_URL=https://your-deployment.crewai.com
heroku config:set CREWAI_BEARER_TOKEN=your-token

Your webhook URL is https://<your-app>.herokuapp.com/webhook/messages — set this as WEBHOOK_URL in AMP.

3. Connecting the two

Once both are deployed:

1. Copy your Heroku app URL (e.g., https://my-app.herokuapp.com)
2. In AMP, set WEBHOOK_URL=https://my-app.herokuapp.com/webhook/messages
3. Redeploy the flow on AMP so it picks up the new env var

API Endpoints

Browser-facing

Endpoint	Method	Description
/api/warmup	POST	Pings AMP /inputs to warm up the deployment
/api/kickoff	POST	Receives job URL + resume PDF, forwards to AMP, returns session_id and kickoff_id
/api/stream/<session_id>	GET	SSE stream — pushes real-time step updates and final report
/api/status/<kickoff_id>	GET	Fallback polling endpoint — proxies to AMP status API

Webhook (called by the flow's event listener)

Endpoint	Method	Description
/webhook/messages	POST	Receives step updates and final report from WebhookEventListener

Report Output

The generated report follows this structure:

• Position — Company and job title
• Candidate — Name only (PII redacted)
• Required Skills — Full list from the posting
• Fitness Score — X/100 with interpretation
• Strengths — Matched skills
• Gaps / Missing Skills — Unmatched requirements
• Summary — 2-3 sentence overall assessment