# mithunraj-m/smart_resume_screener architecture

> Architecture diagram of the GitHub repository mithunraj-m/smart_resume_screener, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-03). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/mithunraj-m/smart_resume_screener
- Repository: https://github.com/mithunraj-m/smart_resume_screener

## Overview

Smart Resume Screener accepts a job-description PDF and one or more resume PDFs, then returns ranked candidates with category scores and an LLM-written rationale. The Streamlit UI submits files to the FastAPI backend; the LangGraph workflow extracts and structures resume content, translates job requirements, matches and scores categories, and generates a summary. Gemini, the embedding model, and Pinecone support analysis and storage. The README documents these capabilities; some internals, especially resume chunking and vector search details, were not fully excerpted.

Main entry points: [`resume_screener/backend/process_resume.py`](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/process_resume.py) (Resume processing), [`resume_screener/backend/process_jd.py`](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/process_jd.py) (JD translation and scoring), [`resume_screener/backend/main.py`](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/main.py) (FastAPI endpoints), [`resume_screener/backend/graph.py`](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/graph.py) (LangGraph workflow)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_interface["User interface"]
  node_ui["Streamlit interface<br/>[app.py]"]
end

subgraph group_api["API orchestration"]
  node_api_entry["FastAPI endpoints<br/>[main.py]"]
end

subgraph group_workflow["Analysis workflow"]
  node_pdf_text["PDF text extraction<br/>[process_resume.py]"]
  node_graph["LangGraph workflow<br/>[graph.py]"]
  node_resume_process["Resume processing<br/>[process_resume.py]"]
  node_jd_process["JD translation and scoring<br/>[process_jd.py]"]
  node_summary["Summary generation"]
  node_state["Workflow state<br/>[state.py]"]
end

subgraph group_services["AI and storage"]
  node_gemini["Gemini LLM"]
  node_embeddings["Sentence embeddings"]
  node_pinecone[("Pinecone index")]
end

node_user(("Recruiter"))

node_user -->|"uploads PDFs"| node_ui
node_ui -->|"posts files"| node_api_entry
node_api_entry -->|"extracts JD text"| node_pdf_text
node_api_entry -->|"streams analysis"| node_graph
node_api_entry -->|"translates batch JD"| node_jd_process
node_graph -->|"runs first"| node_resume_process
node_graph -->|"runs next"| node_jd_process
node_graph -->|"runs last"| node_summary
node_resume_process -->|"extracts resume text"| node_pdf_text
node_resume_process -->|"extracts sections"| node_gemini
node_jd_process -->|"structures requirements"| node_gemini
node_summary -->|"generates rationale"| node_gemini
node_resume_process -.->|"embeds chunks"| node_embeddings
node_jd_process -->|"embeds queries"| node_embeddings
node_resume_process -->|"stores vectors"| node_pinecone
node_resume_process -.->|"updates workflow data"| node_state
node_jd_process -->|"updates scores"| node_state
node_summary -->|"writes summary"| node_state

click node_ui "https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/frontend/app.py"
click node_api_entry "https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/main.py"
click node_pdf_text "https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/process_resume.py"
click node_graph "https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/graph.py"
click node_resume_process "https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/process_resume.py"
click node_jd_process "https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/process_jd.py"
click node_summary "https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/generate_summary.py"
click node_state "https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/state.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_ui toneBlue
class node_api_entry toneAmber
class node_pdf_text,node_graph,node_resume_process,node_jd_process,node_summary,node_state toneMint
class node_gemini,node_embeddings,node_pinecone toneRose
class node_user toneIndigo
```

## Components

### User interface

- [Streamlit interface](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/frontend/app.py): `resume_screener/frontend/app.py`

### API orchestration

- [FastAPI endpoints](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/main.py): `resume_screener/backend/main.py`

### Analysis workflow

- [PDF text extraction](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/process_resume.py): `resume_screener/backend/process_resume.py`
- [LangGraph workflow](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/graph.py): `resume_screener/backend/graph.py`
- [Resume processing](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/process_resume.py): `resume_screener/backend/process_resume.py`
- [JD translation and scoring](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/process_jd.py): `resume_screener/backend/process_jd.py`
- [Summary generation](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/generate_summary.py): `resume_screener/backend/generate_summary.py`
- [Workflow state](https://github.com/mithunraj-m/smart_resume_screener/blob/main/resume_screener/backend/state.py): `resume_screener/backend/state.py`

### AI and storage

- **Gemini LLM**
- **Sentence embeddings**
- **Pinecone index**

### Other components

- **Recruiter**

## Connections

- Recruiter → Streamlit interface: uploads PDFs (evidence: `README.md`)
- Streamlit interface → FastAPI endpoints: posts files (evidence: `resume_screener/frontend/app.py`)
- FastAPI endpoints → PDF text extraction: extracts JD text (evidence: `resume_screener/backend/main.py`)
- FastAPI endpoints → LangGraph workflow: streams analysis (evidence: `resume_screener/backend/main.py`)
- FastAPI endpoints → JD translation and scoring: translates batch JD (evidence: `resume_screener/backend/main.py`)
- LangGraph workflow → Resume processing: runs first (evidence: `resume_screener/backend/graph.py`)
- LangGraph workflow → JD translation and scoring: runs next (evidence: `resume_screener/backend/graph.py`)
- LangGraph workflow → Summary generation: runs last (evidence: `resume_screener/backend/graph.py`)
- Resume processing → PDF text extraction: extracts resume text (evidence: `resume_screener/backend/process_resume.py`)
- Resume processing → Gemini LLM: extracts sections (evidence: `resume_screener/backend/process_resume.py`)
- JD translation and scoring → Gemini LLM: structures requirements (evidence: `resume_screener/backend/process_jd.py`)
- Summary generation → Gemini LLM: generates rationale (evidence: `resume_screener/backend/generate_summary.py`)
- Resume processing → Sentence embeddings: embeds chunks (evidence: `README.md`)
- JD translation and scoring → Sentence embeddings: embeds queries (evidence: `resume_screener/backend/process_jd.py`)
- Resume processing → Pinecone index: stores vectors (evidence: `resume_screener/backend/process_resume.py`)
- Resume processing → Workflow state: updates workflow data (evidence: `resume_screener/backend/process_resume.py`)
- JD translation and scoring → Workflow state: updates scores (evidence: `resume_screener/backend/process_jd.py`)
- Summary generation → Workflow state: writes summary (evidence: `resume_screener/backend/generate_summary.py`)
