JustHireMe

Local-first AI job intelligence for scraping better roles, ranking fit, and generating tailored application materials.

What It Does · Workflow · Architecture · Quick Start · Standard Procedure · Agent Skill + MCP · Contributing · Roadmap

Star History

The Short Version

JustHireMe is an AGPL-licensed, local-first desktop workbench for people who are tired of noisy job boards and black-box AI apply tools.

Maintainer

JustHireMe is built and maintained by Vasudev Siddh - full-stack AI engineer, open-source builder, and the person who got tired of bad job boards and built something better.

Sponsor the project - Sponsorship keeps JustHireMe actively maintained, funds source adapter coverage, and supports the local-first architecture that makes this different from every cloud-first alternative.

Current Status

JustHireMe's stable core is the local-first desktop workbench, Python sidecar API, lead ingestion, deterministic ranking, profile-aware matching, local CRM workflows, and document/outreach generation. Every release is built by GitHub Actions from a v* tag and now ships installers for Windows, macOS, and Linux.

If you are new here, start with the frontend preview first. If you want to contribute backend behavior, source adapters, or packaging, use the full desktop setup below.

It helps you:

Browser automation and auto-apply code exists in the repository, but it is experimental, opt-in, and unsupported as part of the stable core. The supported open-source core is scraper, ranker, vector matching, and customizer.

Visual Workflow

flowchart LR
    A["Import Resume / Profile"] --> B["Profile Graph<br/>Kuzu"]
    A --> C["Skill + Project Vectors<br/>LanceDB"]
    D["Scraper Sources"] --> E["Normalized Leads"]
    E --> F["Lead Quality Gate"]
    F --> G["Ranker + Evaluator"]
    B --> G
    C --> G
    G --> H["Lead Pipeline"]
    H --> I["Customizer"]
    I --> J["Resume PDF"]
    I --> K["Cover Letter PDF"]
    I --> L["Outreach Drafts"]

flowchart TD
    S["Source Adapter"] --> N["Normalize Lead"]
    N --> Q{"Quality Gate"}
    Q -- "Reject" --> R["Filtered with reason"]
    Q -- "Accept" --> C["Local CRM"]
    C --> E["Fit Evaluation"]
    E --> U["User Review"]

    Q:::gate
    R:::reject
    C:::store
    E:::score

    classDef gate fill:#e0f2fe,stroke:#0284c7,color:#0f172a
    classDef reject fill:#fee2e2,stroke:#dc2626,color:#0f172a
    classDef store fill:#dcfce7,stroke:#16a34a,color:#0f172a
    classDef score fill:#fef3c7,stroke:#d97706,color:#0f172a

What It Does

Why This Exists

Most job search tools make one of two mistakes:

JustHireMe takes a different path:

More signal.
More explanation.
More local control.
More contributor-friendly source adapters.
Less blind automation.

Product Principles

Architecture

flowchart TB
    subgraph Desktop["Tauri Desktop Shell"]
        UI["React + TypeScript UI"]
        Bridge["Tauri Commands<br/>Port + API Token"]
    end

    subgraph Backend["Python Backend Sidecar"]
        API["FastAPI + WebSockets"]
        Scrapers["Scraper Agents"]
        Gate["Lead Quality Gate"]
        Ranker["Ranker / Evaluator"]
        Customizer["Document + Outreach Generator"]
    end

    subgraph LocalData["Local Data Layer"]
        SQLite["SQLite CRM"]
        Kuzu["Kuzu Profile Graph"]
        LanceDB["LanceDB Vectors"]
        Files["Generated PDFs"]
    end

    UI --> Bridge --> API
    API --> Scrapers --> Gate --> SQLite
    API --> Ranker
    Ranker --> Kuzu
    Ranker --> LanceDB
    API --> Customizer --> Files
    API --> SQLite

First-run runtime pack. To keep installers small (~100 MB instead of ~450-700 MB), the heavy runtime - the Playwright browser, vector libraries, and the ONNX embedding model - is not bundled into the installer. The app downloads it once on first run, verifies it, and caches it. The pack is content-versioned, so routine app updates reuse the cached pack instead of re-downloading it.

More detail: docs/ARCHITECTURE.md

Repository Map

JustHireMe/
|-- src/                         React frontend workbench
|   |-- api/                     HTTP/WebSocket API clients and types
|   |-- features/                Activity, apply, dashboard, graph, inbox, pipeline, profile, settings
|   |-- shared/                  Components, hooks, context, utilities
|   `-- main.tsx                 Frontend entrypoint
|-- backend/                     Python FastAPI sidecar
|   |-- api/                     FastAPI app, routers, auth, scheduler, WebSockets
|   |-- automation/              Scout scraping orchestration + experimental apply/actuator
|   |-- core/                    Config, occupation vocab, SSRF guard, telemetry
|   |-- data/                    SQLite repos, Kùzu graph, LanceDB vectors, migrations
|   |-- discovery/               Query generation, dedup, quality gate, source adapters
|   |-- gateway/                 Async job tracking, discovery config
|   |-- generation/              Resume, cover letter, and outreach generators
|   |-- graph_service/           Query helpers behind the knowledge-graph view
|   |-- llm/                     Provider abstraction (keyed + keyless/subscription-CLI), embeddings
|   |-- profile/                 Resume/LinkedIn/GitHub/portfolio ingestion + normalization
|   |-- ranking/                 Fit scoring, criteria, semantic/evaluator logic
|   `-- tests/                   Backend unit, integration, and regression tests
|-- src-tauri/                   Tauri Rust shell
|   `-- resources/               Bundled sidecar/runtime resources generated for packages
|-- website/                     Public website
|-- docs/                        Architecture, source adapter, release docs
|-- scripts/                     Build scripts
|-- skills/                      Agent skill instructions
`-- .github/                    CI, issue templates, PR template

Quick Start

Install The Desktop App

Use this path if you are not a developer and just want to run JustHireMe. Open the latest GitHub Release and grab the asset for your platform.