Skip to content

HardMax71/ResuMariner

BranchesTags

Repository files navigation

ResuMariner

We are looking for backend Python Developer for parsing project with unknown bounds and requirements. By the way, have you worked with RAGs and vector databases?

— Undisclosed CTO during what was supposed to be a casual coffee chat

This project started after one of those classic meetings where "simple resume parsing" mysteriously evolved into "enterprise-grade candidate matching with AI predictions." Instead of building yet another janky one-off script, I figured why not create a proper architecture that can actually handle whatever random requirements get tossed in next week? So here's ResuMariner - a solid microservice setup with Neo4j, Qdrant, and various APIs for handling CVs that can grow from "just extract the text" to "find me candidates with 7+ years React who once thought about learning Rust and might enjoy ping-pong" without needing a complete rewrite.

✨ Features

  • CV Upload and Processing: Upload CV documents (PDF, DOCX, JPG) for automated parsing and analysis
  • Structured Data Extraction: Extract key information from CVs into a structured format
  • CV Review Generation: Generate professional reviews of candidate CVs
  • Semantic Search: Find candidates based on natural language queries
  • Structured Search: Search candidates by specific criteria (skills, experience, etc.)
  • Hybrid Search: Combine semantic and structured search for optimal results
  • Secure Architecture: Separation between public-facing and internal services

🏗️ Architecture

ResuMariner follows a microservices architecture with the following components:

sys_diagram

  • Traefik: Reverse proxy and API gateway
  • cv-intake-service: Handles CV file uploads and job management
  • cv-processing-service: Processes and analyzes CV documents using LLMs
  • cv-storage-service: Manages persistent storage of CV data
  • cv-search-service: Provides advanced search capabilities
  • Databases:
    • Neo4j: Graph database for structured CV data
    • Qdrant: Vector database for semantic search embeddings
    • Redis: Job queue and caching

The system is split between a public network zone (accessible from outside) and a protected internal network zone.

🚀 Getting Started

Prerequisites

  • Docker and Docker Compose
  • At least 4GB of RAM for running all services
  • API keys for LLM providers (OpenAI, Gemini, etc.) if using those services (only one is enough)

Deployment

Deployment is straightforward:

docker-compose up --build

That's it! The system will be available at the following endpoints:

📋 API Documentation

Upload API (intake.localhost)

  • POST /api/v1/upload: Upload a CV file for processing
  • GET /api/v1/status/{job_id}: Check processing status
  • GET /api/v1/results/{job_id}: Get processing results

Search API (search.localhost)

  • POST /search/semantic: Semantic search using natural language
  • POST /search/structured: Structured search using specific criteria
  • POST /search/hybrid: Combined semantic and structured search
  • GET /filters: Get available filter options

🧪 Testing

The system includes a comprehensive testing script that can process a CV and run various searches against it:

python test_script.py path/to/resume.pdf

Calling

python test_script.py

Will return result for existing CV in ./test_inputs folder, namely ./test_inputs/Max_Azatian.pdf.

Additional flags are to be found in test_script.py.

Example output for provided CV 
/Users/test/PycharmProjects/pythonProject/venv/bin/python /Users/test/PycharmProjects/ResuMariner/test_script.py 
Using intake service at: http://intake.localhost
Using search service at: http://search.localhost
No review?: False
No store?: False
No search?: False
Parallel?: False
Full JSON?: False
Processing file: ./test_inputs/Max_Azatian_CV.pdf

1. Uploading file to intake service...
Sending to: http://intake.localhost/api/v1/upload
Upload successful. Job ID: 730b6563-e5c3-476d-b535-d46b56a6ed32

Waiting for processing to complete...
 22%|██▏       | 13/60 [00:26<01:35,  2.03s/it]
Completed! Contents: {'job_id': '730b6563-e5c3-476d-b535-d46b56a6ed32', 'status': 'completed', 'created_at': '2025-03-02T23:05:18.196529', 'updated_at': '2025-03-02T23:05:43.193148', 'result_url': '/api/v1/results/730b6563-e5c3-476d-b535-d46b56a6ed32', 'error': ''}
Fetching result from: http://intake.localhost/api/v1/results/730b6563-e5c3-476d-b535-d46b56a6ed32
Processing completed! CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650

2. CV Summary:
================================================================================
{
  "personal_info": {
    "name": "Max Azatian",
    "resume_lang": "en",
    "contact": {
      "email": "max.azatian@gmail.com",
      "phone": "tel:+49-178-1234567",
      "links": {
        "telegram": null,
        "linkedin": "https://www.linkedin.com/in/max-azatian/",
        "github": "https://github.com/HardMax71",
        "other_links": null
      }
    },
    "demographics": {
      "current_location": {
        "city": "Munich",
        "state": null,
        "country": "Germany"
      },
      "work_authorization": {
        "citizenship": "Germany",
        "work_permit": null,
        "visa_sponsorship_required": null
      }
    }
  },
  "professional_profile": {
    "summary": "Backend developer with experience in crafting efficient systems. Combine technical precision with teamwork to solve problems through a clean, maintainable design.",
    "preferences": {
      "role": "Software Engineer",
      "employment_types": [
        "full-time",
        "part-time",
        "contract"
      ],
      "work_modes": [
        "remote",
        "onsite",
        "hybrid"
      ],
      "salary": null
    }
  },
  "skills": [
    "Python",
    "JavaScript",
    "HTML/CSS",
    "FastAPI",
    "Flask",
    "Django",
    "ORM (SQLAlchemy)",
    "Linters (Flake8, Ruff)",
    "Locust",
    "RESTful APIs",
    "PostgreSQL",
    "MongoDB",
    "SQLite",
    "Redis",
    "Neo4j",
    "Docker",
    "Docker Compose",
    "Kubernetes",
    "Git",
    "Linux",
    "Cloud platforms (GCP, Azure)",
    "Traefik",
    "Prometheus",
    "Grafana"
  ],
  "employment_history": [
    {
      "company": "Self-employed",
      "company_url": null,
      "position": "Software Engineer (Part-time)",
      "employment_type": "part-time",
      "work_mode": "remote",
      "duration": {
        "date_format": "MM.YYYY",
        "start": "06.2021",
        "end": "current",
        "duration_months": 45
      },
      "location": {
        "city": "Remote",
        "state": null,
        "country": null
      },
      "key_points": [
        "Reached 90% pilot user adoption for pet salon booking system by developing Django/PostgreSQL MVP with auto scheduling, accompanied by Doxygen documentation featuring UML diagrams (class, activity, deployment)",
        "Increased monthly active user growth by 40% for board game community platform via creating Django REST backend implementation supporting profile creation, blog posts, and filtered game searches with Bootstrap-based UI."
      ],
      "tech_stack": [
        "Django",
        "PostgreSQL"
      ]
    }
  ],
  "projects": [
    {
      "title": "Integr8sCode",
      "url": "https://github.com/HardMax71/Integr8sCode",
      "tech_stack": [
        "Python",
        "FastAPI",
        "Pydantic",
        "Kubernetes",
        "Svelte",
        "MongoDB",
        "Prometheus",
        "Grafana"
      ],
      "key_points": [
        "Reduced memory usage by 25%, tracked in Prometheus, by enforcing per-pod CPU/memory limits (CPU: 100m, memory: 128Mi), adding auto-scaling policies (Horizontal Pod Autoscaler), and optimizing Docker layers (e.g., moving nonessential packages and using python:{version}-slim)",
        "Achieved 30K+ daily script executions with an error rate below 0.3%, monitored via Grafana, by orchestrating ephemeral K8s pods and adding request validation with Pydantic",
        "Raised backend test coverage to 92%, as measured by Codecov, by introducing unit/integration tests for critical modules and logging/reporting coverage in HTML reports."
      ]
    },
    {
      "title": "Flet-Chat",
      "url": "https://github.com/HardMax71/Flet-Chat",
      "tech_stack": [
        "Python",
        "FastAPI",
        "Flet",
        "PostgreSQL",
        "Redis",
        "Docker",
        "Pydantic"
      ],
      "key_points": [
        "Created real-time chat with an average latency of less than 100ms, confirmed by stress tests, by integrating FastAPI WebSockets and Flet UI components",
        "Acquired \u201dA\u201d code quality (measured by Codacy) and 85% test coverage, validated via Codecov, by standardizing linting rules and expanding unit/integration tests",
        "Reduced database load by 31.2%, tracked in Docker, by adopting Redis for caching ephemeral data."
      ]
    },
    {
      "title": "ResuMariner",
      "url": "https://github.com/HardMax71/ResuMariner",
      "tech_stack": [
        "Python",
        "FastAPI",
        "Pydantic",
        "Neo4j",
        "Qdrant",
        "Docker",
        "Traefik",
        "Redis",
        "LLM"
      ],
      "key_points": [
        "Increased data retrieval accuracy by 78% compared to keyword-based SQL queries via implementing hybrid search (Neo4j graph traversals, Qdrant vector similarity), validated through precision/recall on technical skill matching",
        "Lowered mean processing time from 9s to 2.5s by creating parallel processing pipelines and LLM prompting strategies (context windowing, structured output schemas)",
        "Added zero public exposure of internal services, confirmed via OWASP-based penetration testing, by implementing network isolation with Traefik\u2019s internal entrypoints and dedicated bridge networks."
      ]
    }
  ],
  "education": [
    {
      "institution": "Technical University of Munich",
      "qualification": "Master",
      "study_field": "Computer Science",
      "location": {
        "city": "Munich",
        "state": null,
        "country": "Germany"
      },
      "start": "10.2024",
      "end": "current",
      "status": "ongoing",
      "coursework": null,
      "extras": null
    },
    {
      "institution": "Technical University of Munich",
      "qualification": "Bachelor",
      "study_field": "Computer Science",
      "location": {
        "city": "Munich",
        "state": null,
        "country": "Germany"
      },
      "start": "10.2019",
      "end": "08.2024",
      "status": "completed",
      "coursework": null,
      "extras": null
    }
  ],
  "courses": null,
  "certifications": null,
  "language_proficiency": [
    {
      "language": "English",
      "level": {
        "self_assessed": "C1",
        "cefr": "C1"
      }
    },
    {
      "language": "German",
      "level": {
        "self_assessed": "C1",
        "cefr": "C1"
      }
    },
    {
      "language": "Japanese",
      "level": {
        "self_assessed": "B1",
        "cefr": "B1"
      }
    }
  ],
  "awards": null,
  "scientific_contributions": null,
  "validation_metadata": {
    "text_characters_processed": 2808,
    "links_processed": 6,
    "anomalies": []
  }
}
================================================================================

CV Review:
================================================================================
{
  "personal_info": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": null
  },
  "professional_profile": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": null
  },
  "skills": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": null
  },
  "employment_history": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": "Consider adding more quantifiable achievements to further demonstrate impact."
  },
  "projects": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": null
  },
  "education": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": "Consider highlighting relevant coursework or thesis for the ongoing Master's program."
  },
  "courses": {
    "MUST": "Consider adding any relevant courses taken, including names, organizations, and completion years.",
    "SHOULD": null,
    "ADVISE": null
  },
  "certifications": {
    "MUST": "Consider adding any relevant certifications, including issuing organizations and dates.",
    "SHOULD": null,
    "ADVISE": null
  },
  "language_proficiency": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": null
  },
  "awards": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": null
  },
  "scientific_contributions": {
    "MUST": null,
    "SHOULD": null,
    "ADVISE": null
  }
}
================================================================================

3. Running comprehensive search testing

Extracted search parameters:
{
  "name": "Max Azatian",
  "email": "max.azatian@gmail.com",
  "all_skills": [
    "Python",
    "JavaScript",
    "HTML/CSS",
    "FastAPI",
    "Flask",
    "Django",
    "ORM (SQLAlchemy)",
    "Linters (Flake8, Ruff)",
    "Locust",
    "RESTful APIs",
    "PostgreSQL",
    "MongoDB",
    "SQLite",
    "Redis",
    "Neo4j",
    "Docker",
    "Docker Compose",
    "Kubernetes",
    "Git",
    "Linux",
    "Cloud platforms (GCP, Azure)",
    "Traefik",
    "Prometheus",
    "Grafana"
  ],
  "top_skills": [
    "Python",
    "JavaScript",
    "HTML/CSS"
  ],
  "role": "Software Engineer",
  "city": "Munich",
  "country": "Germany",
  "location": "Munich, Germany",
  "companies": [
    "Self-employed"
  ],
  "latest_company": "Self-employed",
  "technologies": [
    "Traefik",
    "Qdrant",
    "LLM",
    "Pydantic",
    "PostgreSQL",
    "Neo4j",
    "Kubernetes",
    "Flet",
    "FastAPI",
    "MongoDB",
    "Django",
    "Docker",
    "Python",
    "Grafana",
    "Redis",
    "Prometheus",
    "Svelte"
  ],
  "top_technologies": [
    "Traefik",
    "Qdrant",
    "LLM"
  ]
}

Running Semantic search by name: Max Azatian
Sending to: http://search.localhost/search/semantic
Payload: {
  "query": "Max Azatian",
  "limit": 5,
  "min_score": 0.0
}
Found 1 results (semantic search):

Search Results (Semantic search by name):
Parameters: Max Azatian
================================================================================
1. CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 1.0000
   Number of matches: 5
----------------------------------------
================================================================================

Running Semantic search by role: Software Engineer
Sending to: http://search.localhost/search/semantic
Payload: {
  "query": "Software Engineer",
  "limit": 5,
  "min_score": 0.0
}
Found 1 results (semantic search):

Search Results (Semantic search by role):
Parameters: Software Engineer
================================================================================
1. CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 1.0000
   Number of matches: 5
----------------------------------------
================================================================================

Running Semantic search by company: Self-employed
Sending to: http://search.localhost/search/semantic
Payload: {
  "query": "Self-employed",
  "limit": 5,
  "min_score": 0.0
}
Found 1 results (semantic search):

Search Results (Semantic search by company):
Parameters: Self-employed
================================================================================
1. CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 0.6046
   Number of matches: 5
----------------------------------------
================================================================================

Running Semantic search by location: Munich, Germany
Sending to: http://search.localhost/search/semantic
Payload: {
  "query": "Munich, Germany",
  "limit": 5,
  "min_score": 0.0
}
Found 1 results (semantic search):

Search Results (Semantic search by location):
Parameters: Munich, Germany
================================================================================
1. CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 1.0000
   Number of matches: 5
----------------------------------------
================================================================================

Running Semantic search by skills: Python, JavaScript, HTML/CSS
Sending to: http://search.localhost/search/semantic
Payload: {
  "query": "Python, JavaScript, HTML/CSS",
  "limit": 5,
  "min_score": 0.0
}
Found 1 results (semantic search):

Search Results (Semantic search by skills):
Parameters: Python, JavaScript, HTML/CSS
================================================================================
1. CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 0.4848
   Number of matches: 5
----------------------------------------
================================================================================

Running Structured search by top skills: skills=[Python, JavaScript, HTML/CSS]
Sending to: http://search.localhost/search/structured
Payload: {
  "limit": 5,
  "skills": [
    "Python",
    "JavaScript",
    "HTML/CSS"
  ]
}
Found 1 results (structured search):

Search Results (Structured search by top skills):
Parameters: skills=[Python, JavaScript, HTML/CSS]
================================================================================
1. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 1.0000
   Number of matches: 0
----------------------------------------
================================================================================

Running Structured search by role: role=Software Engineer
Sending to: http://search.localhost/search/structured
Payload: {
  "limit": 5,
  "role": "Software Engineer"
}
Found 1 results (structured search):

Search Results (Structured search by role):
Parameters: role=Software Engineer
================================================================================
1. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 1.0000
   Number of matches: 0
----------------------------------------
================================================================================

Running Structured search by company: company=Self-employed
Sending to: http://search.localhost/search/structured
Payload: {
  "limit": 5,
  "company": "Self-employed"
}
Found 1 results (structured search):

Search Results (Structured search by company):
Parameters: company=Self-employed
================================================================================
1. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 1.0000
   Number of matches: 0
----------------------------------------
================================================================================

Running Structured search by skills and role: skills=[Python, JavaScript, HTML/CSS], role=Software Engineer
Sending to: http://search.localhost/search/structured
Payload: {
  "limit": 5,
  "skills": [
    "Python",
    "JavaScript",
    "HTML/CSS"
  ],
  "role": "Software Engineer"
}
Found 1 results (structured search):

Search Results (Structured search by skills and role):
Parameters: skills=[Python, JavaScript, HTML/CSS], role=Software Engineer
================================================================================
1. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 1.0000
   Number of matches: 0
----------------------------------------
================================================================================

Running Structured search by technologies: technologies=[Traefik, Qdrant, LLM]
Sending to: http://search.localhost/search/structured
Payload: {
  "limit": 5,
  "technologies": [
    "Traefik",
    "Qdrant",
    "LLM"
  ]
}
Found 1 results (structured search):

Search Results (Structured search by technologies):
Parameters: technologies=[Traefik, Qdrant, LLM]
================================================================================
1. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 1.0000
   Number of matches: 0
----------------------------------------
================================================================================

Running Hybrid search by name: query=Max Azatian
Sending to: http://search.localhost/search/hybrid
Payload: {
  "query": "Max Azatian",
  "limit": 5,
  "vector_weight": 0.7,
  "graph_weight": 0.3
}
Found 2 results (hybrid search):

Search Results (Hybrid search by name):
Parameters: query=Max Azatian
================================================================================
1. CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 0.7000
   Number of matches: 10
----------------------------------------
2. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 0.3000
   Number of matches: 0
----------------------------------------
================================================================================

Running Hybrid search by name with skills: query=Max Azatian, skills=[Python, JavaScript, HTML/CSS]
Sending to: http://search.localhost/search/hybrid
Payload: {
  "query": "Max Azatian",
  "limit": 5,
  "vector_weight": 0.7,
  "graph_weight": 0.3,
  "skills": [
    "Python",
    "JavaScript",
    "HTML/CSS"
  ]
}
Found 1 results (hybrid search):

Search Results (Hybrid search by name with skills):
Parameters: query=Max Azatian, skills=[Python, JavaScript, HTML/CSS]
================================================================================
1. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 0.3000
   Number of matches: 0
----------------------------------------
================================================================================

Running Hybrid search by name with technologies: query=Max Azatian, technologies=[Traefik, Qdrant, LLM]
Sending to: http://search.localhost/search/hybrid
Payload: {
  "query": "Max Azatian",
  "limit": 5,
  "vector_weight": 0.7,
  "graph_weight": 0.3,
  "technologies": [
    "Traefik",
    "Qdrant",
    "LLM"
  ]
}
Found 2 results (hybrid search):

Search Results (Hybrid search by name with technologies):
Parameters: query=Max Azatian, technologies=[Traefik, Qdrant, LLM]
================================================================================
1. CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 0.7000
   Number of matches: 10
----------------------------------------
2. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 0.3000
   Number of matches: 0
----------------------------------------
================================================================================

Running Hybrid search by role and company: query=Software Engineer, company=Self-employed
Sending to: http://search.localhost/search/hybrid
Payload: {
  "query": "Software Engineer",
  "limit": 5,
  "vector_weight": 0.7,
  "graph_weight": 0.3,
  "company": "Self-employed"
}
Found 1 results (hybrid search):

Search Results (Hybrid search by role and company):
Parameters: query=Software Engineer, company=Self-employed
================================================================================
1. CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
   Name: Max Azatian
   Email: max.azatian@gmail.com
   Score: 0.3000
   Number of matches: 0
----------------------------------------
================================================================================

4. Search Results Comparison
================================================================================
Total unique CVs found: 2
Search methods: 14

Results by search method:
  semantic_name: 1 results
  semantic_role: 1 results
  semantic_latest_company: 1 results
  semantic_location: 1 results
  semantic_top_skills: 1 results
  structured_top skills: 1 results
  structured_role: 1 results
  structured_company: 1 results
  structured_skills and role: 1 results
  structured_technologies: 1 results
  hybrid_name: 2 results
  hybrid_name with skills: 1 results
  hybrid_name with technologies: 2 results
  hybrid_role and company: 1 results

CVs found by multiple methods:
  CV ID: 730b6563-e5c3-476d-b535-d46b56a6ed32
    Found by 9 methods: structured_top skills, structured_role, structured_company, structured_skills and role, structured_technologies, hybrid_name, hybrid_name with skills, hybrid_name with technologies, hybrid_role and company
  CV ID: f0ffd328-ed4c-4e80-8b77-f4ab1b7b6650
    Found by 7 methods: semantic_name, semantic_role, semantic_latest_company, semantic_location, semantic_top_skills, hybrid_name, hybrid_name with technologies
================================================================================

Test completed successfully!

Process finished with exit code 0

📖 License

This project is licensed under the MIT License - see the LICENSE file for details.

About

CV parsing & search system

Topics

python docker redis microservices neo4j docker-compose graph-database traefik semantic-search resume-parsing fastapi vector-database qdrant

Resources

Readme

License

MIT license
Activity

Stars

0 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages