Live Application | GitHub Repository
Project Overview
ApplicAI is a production-grade AI platform that streamlines the job search process by generating tailored CVs, cover letters, and LinkedIn messages. Evolving from a local Python script to a SaaS serving ~100 daily users, it leverages LLMs to parse documents and create structured application materials, achieving a 10% conversion rate.
Key Features
- Intelligent Parsing: Uses AI to extract structured data from PDF/DOCX resumes into standardized formats.
- Multi-Model AI Support: Integrates Google Gemini and Meta Llama to balance cost and performance.
- Application Tracking: Visualizes job search progress with Sankey diagrams, heatmaps, and analytics.
- Privacy-First Design: Supports a “Bring Your Own Key” (BYOK) model to ensure user data security.
Challenges & Architecture
Designed as a decoupled web app, ApplicAI features a SvelteKit frontend for a reactive UI and a high-performance FastAPI backend. I implemented an end-to-end type safety system where TypeScript definitions are automatically generated from Pydantic models. The infrastructure runs on a self-hosted Hetzner VPS managed via Coolify, with S3-compatible storage for documents and an async architecture tailored for handling concurrent AI stream processing. Check out the GitHub Repository for more details on the implementation.