FS Resume Parser Agent Architecture

System Overview

Python backend for SiftHire resume intelligence: parse documents, store in MongoDB, score/match to JDs, generate PDF outputs, and talk to AWS/Zoho as configured.

Architecture Diagram

graph TB
    ResumeApi["app.py / resume_service<br/>API entry"]
    ParserEngine["parser_engine<br/>extract resume text"]
    PdfEngine["pdf_engine / pdf_converter"]
    MongoDatabase[("MongoDB")]
    AwsS3Storage["AWS S3"]
    GptScoring["GPT scoring / matching"]

    ResumeApi -->|"parses uploaded resumes"| ParserEngine
    ResumeApi -->|"generates PDF outputs"| PdfEngine
    ResumeApi -->|"stores candidates / JDs / resumes"| MongoDatabase
    ResumeApi -->|"stores files"| AwsS3Storage
    ResumeApi -->|"JD match + eligibility score"| GptScoring

How to read this diagram

Arrows mean flow / dependency / "talks to", not button clicks. Each arrow label says what happens on that step.

Core Components

Module	Role
parser_engine.py	Document parsing
pdf_engine.py / pdf_converter.py	PDF generation/conversion
db.py	Mongo access
aws_service.py	S3 and AWS helpers
crypto.py	Crypto/helpers for secure fields
templates/	Resume output templates

Fallback behavior

If DOCX to PDF conversion fails, system falls back to direct DOCX parsing (python-docx / docx2txt) and still extracts text/names.

Technology Stack

Layer	Technology
Language	Python
DB	MongoDB
Storage	AWS S3
AI	GPT key in config
PDF	PDFMiner / DocumentEngine settings

Next Steps

• API Reference
• Getting Started
• Overview