resume-jd-analyzer

Resume–JD Match Analyzer

QLoRA Adapter for Llama-3.2-3B-Instruct

Overview

This model is a QLoRA fine-tuned adapter built on top of meta-llama/Llama-3.2-3B-Instruct.

The model analyzes the match between a candidate resume and a job description and returns structured JSON containing:

• match_score
• missing_skills
• resume_improvements
• interview_focus_areas

Training Details

Base Model

• meta-llama/Llama-3.2-3B-Instruct

Fine-Tuning Method

• QLoRA
• PEFT
• TRL SFTTrainer
• 4-bit NF4 quantization

Dataset

Synthetic dataset generated using GPT-4o.

Dataset size:

• 500 total examples
• 400 training examples
• 100 validation examples

Roles included:

• Backend Engineer
• Frontend Engineer
• Full Stack Engineer
• Data Scientist
• ML Engineer
• DevOps Engineer
• Cloud Engineer
• Cybersecurity Analyst
• AI Engineer

Training Configuration

• Epochs: 2
• LoRA Rank (r): 16
• LoRA Alpha: 32
• LoRA Dropout: 0.05
• Optimizer: paged_adamw_8bit
• Max Sequence Length: 1250

Validation Metrics

• Validation Loss: 0.7125
• Mean Token Accuracy: 79.3%

Example Output

{
  "match_score": 82,
  "missing_skills": [
    "Kubernetes"
  ],
  "resume_improvements": [
    "Add Kubernetes project experience"
  ],
  "interview_focus_areas": [
    "Distributed systems",
    "Container orchestration"
  ]
}

Intended Use

Resume screening, candidate evaluation, recruiting workflows, and interview preparation assistance.

Limitations

• Trained on synthetic data.
• Match scores should be treated as recommendations rather than hiring decisions.
• Human review is recommended for production use.