backstory/frontend/public/docs/resume-generation.md

The system follows a carefully designed pipeline with isolated stages to prevent fabrication:

System Architecture Overview

The system uses a pipeline of isolated analysis and generation steps:

1. Stage 1: Isolated Analysis (three sub-stages)

   • 1A: Job Analysis - Extracts requirements from job description only
   • 1B: Candidate Analysis - Catalogs qualifications from resume/context only
   • 1C: Mapping Analysis - Identifies legitimate matches between requirements and qualifications

2. Stage 2: Resume Generation

   • Uses mapping output to create a tailored resume with evidence-based content

3. Stage 3: Verification

   • Performs fact-checking to catch any remaining fabrications

flowchart TD
    subgraph "Stage 1: Isolated Analysis"
        subgraph "Stage 1A: Job Analysis"
            A1[Job Description Input] --> A2[Job Analysis LLM]
            A2 --> A3[Job Requirements JSON]
        end

        subgraph "Stage 1B: Candidate Analysis"
            B1[Resume Input] --> B5[Candidate Analysis LLM]
            B5 --> B4[Candidate Qualifications JSON]
            B2[Candidate Info] --> B3[RAG]
            B3[RAG] --> B2[Candidate Info]
            A3[Job Requirements JSON] --> B3[RAG]
            B3[RAG] --> B5
        end
        
        subgraph "Stage 1C: Mapping Analysis"
            C1[Job Requirements JSON] --> C3[Mapping Analysis LLM]
            C2[Candidate Qualifications JSON] --> C3
            C3 --> C4[Skills Mapping JSON]
        end
    end
    
    subgraph "Stage 2: Resume Generation"
        D1[Skills Mapping JSON] --> D3[Resume Generation LLM]
        D2[Original Resume Reference] --> D3
        D3 --> D4[Tailored Resume Draft]
    end
    
    subgraph "Stage 3: Verification"
        E1[Skills Mapping JSON] --> E2[Original Materials]
        E2 --> E3[Tailored Resume Draft]
        E3 --> E4[Verification LLM]
        E4 --> E5{Verification Check}
        E5 -->|PASS| E6[Approved Resume]
        E5 -->|FAIL| E7[Correction Instructions]
        E7 --> D3
    end
    
    A3 --> C1
    B4 --> C2
    C4 --> D1
    C4 --> E1
    D4 --> E3
    
    style A2 fill:#f9d77e,stroke:#333,stroke-width:2px
    style B5 fill:#f9d77e,stroke:#333,stroke-width:2px
    style C3 fill:#f9d77e,stroke:#333,stroke-width:2px
    style D3 fill:#f9d77e,stroke:#333,stroke-width:2px
    style E4 fill:#f9d77e,stroke:#333,stroke-width:2px
    style E5 fill:#a3e4d7,stroke:#333,stroke-width:2px
    style E6 fill:#aed6f1,stroke:#333,stroke-width:2px
    style E7 fill:#f5b7b1,stroke:#333,stroke-width:2px

Stage 1: Isolated Analysis (three separate sub-stages)

1. Job Analysis: Extracts requirements from just the job description
2. Candidate Analysis: Catalogs qualifications from just the resume/context
3. Mapping Analysis: Identifies legitimate matches between requirements and qualifications

Stage 2: Resume Generation

Creates a tailored resume using only verified information from the mapping

Stage 3: Verification

1. Performs fact-checking to catch any remaining fabrications
2. Corrects issues if needed and re-verifies

Key Anti-Fabrication Mechanisms

The system uses several techniques to prevent fabrication:

• Isolation of Analysis Stages: By analyzing the job and candidate separately, the system prevents the LLM from prematurely creating connections that might lead to fabrication.
• Evidence Requirements: Each qualification included must have explicit evidence from the original materials.
• Conservative Transferability: The system is instructed to be conservative when claiming skills are transferable.
• Verification Layer: A dedicated verification step acts as a safety check to catch any remaining fabrications.
• Strict JSON Structures: Using structured JSON formats ensures information flows properly between stages.

Implementation Details

• Prompt Engineering: Each stage has carefully designed prompts with clear instructions and output formats.
• Error Handling: Comprehensive validation and error handling throughout the pipeline.
• Correction Loop: If verification fails, the system attempts to correct issues and re-verify.
• Traceability: Information in the final resume can be traced back to specific evidence in the original materials.