Resume AI

Your resume, perfectly tailored with AI.

Comment

Resume AI

Inspiration

The inspiration for Resume AI came from a personal and universal pain point in the job search: the tedious, manual process of tailoring a resume for every single job application. I realized that a significant amount of time was being spent just rephrasing duties and highlighting different skills to match keywords in a job description. This repetitive work was a major bottleneck that limited the number of quality applications one could submit. I wanted to build a tool that could automate this entire process, leveraging AI to give job seekers a powerful advantage and help them land interviews more efficiently.

What it does

Resume AI is an intelligent web application that automates resume optimization. A user provides their existing resume (as a PDF) or can fill out the details manually and a job description for a role they're interested in. The application then:

  • Parses and Extracts: Intelligently extracts structured data from the resume, including contact information, skills, work experience, and education, even detecting hyperlinks.
  • Analyzes the Job: Uses an AI model to analyze the job description, identifying the key requirements, technical skills, and responsibilities the employer is looking for.
  • Tailors the Resume: Generates a newly tailored resume that emphasizes the user's most relevant qualifications and rephrases content to align perfectly with the job description.
  • Provides Insights: Offers a match score and highlights areas for improvement, helping the user understand their fit for the role.

The result is a perfectly customized resume, ready for submission, created in a fraction of the time it would take manually.

How we built it

The application was built using a full-stack Python approach, with a focus on modularity and a robust backend.

  • Frontend: Built with Streamlit, allowing rapid development of an interactive and responsive web application. Plotly was used for data visualizations on the dashboard.
  • Backend & AI: Powered by Python, integrating AI capabilities via the OpenRouter API with carefully engineered prompts for job analysis and resume generation.
  • Data Processing: Used PyMuPDF (fitz) for PDF processing, which outperformed alternatives in accurately extracting text, structured data, and hyperlinks.
  • Database: MongoDB (via pymongo) was integrated for persistence, with a fallback to local JSON storage to ensure functionality even without database access.

How GPT-OSS-20B is used

The GPT-OSS-20B model plays a central role in Resume AI, powering the intelligence behind resume tailoring and job analysis:

  • Job Analysis: The model processes raw job descriptions to extract and summarize the role’s responsibilities, technical requirements, and soft skills.
  • Resume Tailoring: Using carefully designed prompts, GPT-OSS-20B rewrites and restructures resume content to align with the identified keywords and expectations in the job description.
  • Skill Gap Detection: By comparing the candidate’s profile against the parsed job requirements, the model highlights missing or underrepresented skills and suggests improvements.
  • Language Quality: The model ensures that the generated resume content is polished, professional, and ATS-friendly, improving readability while maintaining accuracy.

This integration makes Resume AI much more than a keyword matcher—it becomes a true AI-powered assistant that understands both job postings and candidate profiles at a semantic level.

Challenges we ran into

Throughout development, we faced several technical challenges:

  • Streamlit's Execution Model: Resolved errors like StreamlitAPIException and StreamlitDuplicateElementId by learning the execution flow and assigning unique keys to widgets.
  • Accurate PDF Parsing: Initial methods failed, leading us to adopt PyMuPDF and build sophisticated parsing logic for varied resume formats.
  • State Management: Solved issues like UnboundLocalError by restructuring data loading logic for both database and local file scenarios.
  • Dependency Management: Learned the importance of maintaining requirements.txt to prevent issues like ModuleNotFoundError.

Accomplishments that we're proud of

Key achievements include:

  • Intelligent PDF Processor: Accurately parsing resumes, extracting structured data, and preserving hyperlinks.
  • Robust Fallback System: Seamlessly working with or without a database connection.
  • Effective AI Prompt Engineering: Designing prompts that consistently generate high-quality resume content.
  • Polished User Experience: Built an intuitive Streamlit interface despite the complexity behind the scenes.

What we learned

This project provided immense learning opportunities in:

  • Full-Stack Development with Python: Connecting Streamlit (frontend), backend logic, and MongoDB database.
  • Layered Architecture: Abstracting persistence into a SessionManager to decouple UI from storage and enable fallback.
  • Advanced Data Extraction: Structured data extraction from unstructured PDFs.
  • Iterative Problem Solving: Turning every bug and error into a chance to refactor and improve.

What's next for Resume AI

Future plans include:

  • Cover Letter Generation: AI-powered tailored cover letters based on resume and job description.
  • Application Tracker: Dashboard for tracking job applications, statuses, and reminders.
  • AI-Powered Interview Prep: Generate potential interview questions with answering tips.
  • Enhanced Analytics: Provide insights into missing skills and suggest professional development areas.

Built With

Share this project:

Updates