How to Use AI to Understand Gaps in Your Resume and Job Descriptions

At the beginning of my career, when I applied for jobs, I felt there was always a gap between what my resume had and what the job description said. I used to spend many hours staring at my resume, second-guessing whether it reflected my skills correctly or aligned with the Job Description and what recruiters were looking for. Not knowing if my resume would surpass the Applicant Tracking Systems (ATS) drove me to understand what might work in today's job search game.

That's when I realized that there might be many of those who are struggling with such issues. However, people who are constantly up-leveling their skills in software engineering and have gained might find it easier to at least get through the ATS filtering with their experience, specific job requirements, or domain knowledge needed to crack the job. But for someone who's new to software engineering or who knows how to use AI but needs a head start with building an AI (LLM-based) tool for themselves, this solution might help. 

Step Forward: Building the Solution

I used what I've learned to create the AI-powered resume enhancer,  designed to simplify and supercharge the often daunting resume writing process. Here's how I've approached it:

1. Problem Identification: Most resumes fail to sell your skills, lack clarity, or fail to pass ATS filters. Based on my experience, I realized this needs to be solved. 
2. Going Deep into ATS Tech: I researched how ATS parses resumes to identify keywords and skills. Upon carefully reading some of the low-hanging fruits (keywords, gaps in the skills, what the recruiter/hiring manager would read and find interesting), I've identified the baseline items that can be automated with AI. This helped me establish the "logic" behind using AI to get one step closer to getting through a Job application. 
3. Leveraging GPT models: I've started with OpenAI's GPT3.5 Turbo and then GPT4 to analyze resumes with basic prompts. Later, I increased the details it can extract and analyze to interpret job experience and translate it into clear, recruiter-friendly language. Although this is a study to explore what's possible, we can add more models to the code and get varied responses. Later, we can tune the models to get the best insights. 

What Does This Repository Do?

Here are the two major elements that are currently written in the code, which can be customized or more elements as per the job seeker's need. 

Skills Extraction: The defined prompt (below) will identify the technical, functional, and soft skills — something that I struggled to articulate in my resumes before. The input text can be a resume or even a job description they expect a candidate to have. This will probably also help recruiters (if they want to use this kind of prompt) to make it easier for them to categorically view skills. 

prompt = """
        Please analyze the following resume text and extract three categories of skills:
        1. Technical Skills: Hard skills related to specific technologies, tools, or technical knowledge
        2. Functional Skills: Job-specific abilities and domain knowledge
        3. Soft Skills: Interpersonal and behavioral traits

        For each category, provide a comma-separated list of identified skills.
        If a category has no identifiable skills, mark it as "None identified".

        Resume text:
        {text}

        Please format the response as follows:
        Technical Skills: [list of skills]
        Functional Skills: [list of skills]
        Soft Skills: [list of skills]
        """.format(text=resume_text)

resume_text = """
        Led a team of 5 developers to successfully migrate a legacy Java application 
        to Python, implementing CI/CD pipelines using Jenkins and Docker. 
        Improved system performance by 40% through optimization and reduced deployment 
        time by 60%. Conducted weekly team meetings and mentored 2 junior developers.
        """

Technical Skills: Java, Python, Jenkins, Docker, CI/CD
Functional Skills: None identified
Soft Skills: Leadership, Mentorship

Skill Gap Analysis: The defined prompts will take a resume input and job description input to understand the skill gaps and give you a score explaining the gap between your resume and JD. Although this might be a generic comparison, with a larger resume and JD text, I suggest users experiment with various models (with better reasoning capability) and temperatures to get the expected output. 

resume_text = """John Doe
        Software Engineer with 5+ years of experience in big data technologies. 
        Proficient in Hadoop, Spark, and Kafka. 
        Led a team to develop a data processing pipeline that improved data retrieval times by 30%. 
        Experience in designing and implementing scalable data architectures and data models. 
        Strong background in Python and Java, with a focus on data analysis and machine learning applications.
        """

job_description = """As a Data Engineer, you will be responsible for designing, building, and maintaining scalable data pipelines and architectures. 
        You will work with large datasets to ensure data quality and accessibility for analytics and business intelligence. 
        Key responsibilities include developing ETL processes, optimizing data storage solutions, and collaborating with data scientists and analysts to support data-driven decision-making. 
        Required skills include proficiency in SQL, Python, and big data technologies such as Apache Spark and Hadoop. Experience with cloud platforms like AWS or Azure is a plus. 
        Strong problem-solving skills and the ability to work in a fast-paced environment are essential."""

prompt = """
        Perform a detailed skill gap analysis between the candidate's resume and the job description. 
        Follow these specific analytical steps:

        1. Skill Extraction and Quantification:
           - From Job Description: List required skills with their relative importance (Scale 1-5)
           - From Resume: List candidate's skills with their apparent proficiency level (Scale 1-5)
           Base these ratings on:
           - Years of experience
           - Project complexity described
           - Leadership/ownership level
           - Specific achievements mentioned

        2. Gap Analysis:
           - Calculate match scores for each skill area
           - Identify missing critical skills (importance ≥4)
           - Note areas where candidate exceeds requirements
           
        3. Recommendations:
           - Prioritize skill gaps by:
             * Importance to role (1-5)
             * Estimated time to acquire
             * Impact on job performance
           - Suggest specific learning paths or certifications

        Format the response as follows:
        
        SKILL REQUIREMENTS ANALYSIS
        [List skills from JD with importance ratings]

        CANDIDATE SKILL ASSESSMENT
        [List candidate's skills with proficiency ratings]

        GAP ANALYSIS
        [Quantified gaps and matches]

        CRITICAL GAPS
        [List of high-priority missing skills]

        RECOMMENDATIONS
        [Prioritized action items]

        Resume text:
        {resume}

        Job Description:
        {jd}
        """.format(resume=resume_text, jd=job_description)

Skill Gap Analysis:
SKILL REQUIREMENTS ANALYSIS
- SQL (5)
- Python (4)
- Apache Spark (4)
- Hadoop (4)
- AWS/Azure (3)

CANDIDATE SKILL ASSESSMENT
- SQL: 3
- Python: 5
- Apache Spark: 5
- Hadoop: 5
- AWS/Azure: 0

GAP ANALYSIS
- SQL: Candidate lacks proficiency (Gap: -2)
- Python: Candidate exceeds requirement (Match: +1)
- Apache Spark: Candidate meets requirement (Match: +1)
- Hadoop: Candidate meets requirement (Match: +1)
- AWS/Azure: Candidate lacks proficiency (Gap: -3)

CRITICAL GAPS
- SQL
- AWS/Azure

RECOMMENDATIONS
1. Prioritize learning SQL as it is a critical skill for the role (Importance: 5, Time to acquire: Short, Impact: High)
2. Obtain certification in AWS or Azure to improve proficiency in cloud platforms (Importance: 4, Time to acquire: Medium, Impact: Medium)

Insights: The prompts are written in a way that the model is expected to understand your professional story and suggest a way to frame your resume better. 

Trackable Progress: It logs all the interactions with the model in a log file so that you can see exactly how your resume is improving. Which can also be used to check which model is giving you the expected result. 

Code Structure

src/
├── config.py          # Configuration settings
├── logger.py          # Logging setup
├── openai_client.py   # Handles OpenAI API interactions
├── prompt_manager.py  # Manages prompt templates and logic
└── main.py            # Application entry point

Benefits

This code was written with my own experiences in landing a job opportunity, but this can help many different types of job seekers, like

1. Tech professionals: Those who are trying to sell their hard-earned technical expertise with precision. 
2. Career Changers: Job seekers who are looking for a way to reframe their past experiences in a way that's relevant to a new field.
3. Recent Grads: Students who are looking for internships or their first job opportunity but are struggling to get filtered by ATS. 
4. Everyone Striving for Clarity: I guess this can work for anybody who wants to build a resume that speaks directly to the recruiter's needs. 

Takeaway for Job Seekers

If you find it hard to position yourself for the next step in your career, take it from me; it starts with your resume, but it doesn't have to be frustrating. By leveraging AI and ML, you can shift from guesswork to strategic resume mapping to any Job Description you see. This code helps you begin that process, and I'll add more to this repository as we go along. 

Pro Tips

• Tip #1: Run your most important bullet points through multiple analyses to catch all possible skills. Utilize multiple LLM models that you have access to.
• Tip #2: Use the extracted skills to customize your resume for different job postings. 
• Tip #3: Create multiple templates for resumes catering to variations in Job Descriptions
• Tip #4: Keep the logs to track your resume's evolution and LLM Model Responses.

Success Metrics

• Track callback rates before and after
• Monitor ATS success with tools like Jobscan
• Compare skills match with job requirements

Common Pitfalls to Avoid

• Don't overload with technical jargon
• Avoid generic descriptions
• Don't ignore the soft skills