Back to projects
AI2025FeaturedIn development

AI Job Finder — Discovery & Assisted Apply

AI-powered job discovery & application assistant — scrapes job portals, scores fit against a CV, generates tailored PDFs, and assists form filling with a human-in-the-loop apply flow.

GitHub
  • Python
  • FastAPI
  • Playwright
  • CustomTkinter
  • Pydantic
  • ReportLab
  • BeautifulSoup
  • httpx
  • Browser Automation
  • LLM Integration
  • Web Scraping
  • PDF Generation
  • pytest
  • TypeScript / Bun
AI Job Finder — Discovery & Assisted Apply — 1

Overview

What was built

Local desktop system for automated job search and assisted applications.

Architecture splits a scrape → hard filters → AI match → CV/cover PDF → Playwright apply-assist → human confirm pipeline (job-seeker UI) from a FastAPI AI gateway (Free-AI) that drives free web LLMs (DeepSeek, Duck.ai, Qwen) via browser automation—no paid API keys.

Modular Python packages cover core, scrapers, AI, documents, apply, pipeline, and UI.

Portal scrapers include LinkedIn guest API, Remotive, and Arbeitnow with deduplication and persistent job state.

Material-style CustomTkinter dashboard for filters, CV editing, loop control, and document preview.

01

Challenges

Keep multi-portal scraping reliable with deduplication and persistent state; score CV–job fit without paid LLM APIs; automate form filling without blind auto-submit; and separate desktop UI, pipeline, and AI gateway concerns cleanly.

02

Solutions

Modular Python packages with clear SoC; FastAPI gateway with worker pools, sticky sessions, and request queues over free browser LLMs; Playwright for login/session reuse and assisted apply with human confirmation; ReportLab/PyPDF for tailored PDFs; httpx + BeautifulSoup/lxml scrapers; CustomTkinter dashboard; pytest + Ruff for quality.

More work