JobOS

JobOS is a local-first AI job intelligence system designed for precision, control, and privacy. Unlike generic LLM wrappers that blindly rewrite resumes, JobOS treats resumes as a deterministic compilation problem: parsing job requirements, scoring your structured accomplishments against them, and compiling a tailored LaTeX resume that aligns perfectly.

🚀 Current Status

• 🛠 Working on the core Resume Builder.
• 📄 Working on the Deterministic LaTeX Renderer.
• 🛡 Focusing heavily on strictly validating all LLM responses (enforcing JSON schemas, preventing hallucinations, keeping determinism).

✨ Features

• Local-First & Privacy-Focused: No cloud API keys or tracking. Uses a local LLM via Ollama for all AI processes and a local SQLite database for fast, private data storage.
• Structured Resume Inventory: Store your projects, experience, and skills natively as structured entities complete with impact scores, recency metrics, and tags.
• Deterministic Resume Compilation Engine: Ranks inventory items based on a custom weighted scoring function (keyword match, semantic similarity, impact, recency) and selects top candidates under strict sizing and section constraints.
• ATS-Optimized LaTeX Renderer: Assembles the top-ranked modules into a beautiful, standardized LaTeX template and automatically compiles it directly to a PDF using a local LaTeX distribution (like MikTeX).
• Structured LLM Job Parsing: Uses bounded AI processing to precisely convert raw, messy job descriptions into structured data formats (must-haves, nice-to-haves, exact keywords, role type, seniority).
• Match Insights Engine: Provides an actionable decision support system outlining your match score, missing skills, and the strongest signals for a given posting.
• Job Tracker & Browser Extension Integration: Track the application pipeline states (Saved, Applied, OA, Interview, Offer, Rejected) natively and effortlessly parse jobs from the web directly into your JobOS inbox.
• Bounded AI Enhancement: AI is explicitly restricted to bullet clarity, keyword injection, and data extraction. The system enforces generating NO hallucinated content.

🏗 Architecture

• Frontend: Next.js (React) + Tailwind CSS via Tauri
• App Core: Rust (Tauri Core) + Local API Layer
• Database: SQLite
• AI/LLM Engine: local Ollama
• Document Engine: LaTeX Compiler

🛠 Getting Started

Prerequisites

• Node.js >= 18
• Rust (for Tauri setup)
• Ollama (running locally)
• MikTeX or an equivalent full local LaTeX distribution

Running JobOS Locally

1. Install module dependencies:

   npm install

2. Start the development server alongside the Tauri desktop app:

   npm run tauri dev