Loading project...
Back to Work
ApplyFlow
AI-powered job application tracker that turns a pasted job posting into a structured, trackable application — complete with an AI-generated fit assessment against your resume.
Built to solve a real problem from my own job search: manually re-typing job postings into a spreadsheet, then guessing whether a role is actually worth applying to, is tedious and error-prone. ApplyFlow automates both.
Features
Automated Data Extraction
Paste JD → AI extraction. The system parses unstructured job descriptions into a structured schema, capturing role requirements, responsibilities, and metadata.
AI Fit Analysis
AI fit analysis evaluates the extracted job requirements against your stored resume, generating a targeted fit score and highlighting specific strengths and gaps.
Pipeline Tracking
Review/save → pipeline tracking. A centralized dashboard visualizes application states across stages, keeping the job search organized.
Optional Reminders
Optional follow-up reminders can be toggled per application, ensuring you never miss a beat when waiting to hear back from recruiters.
System Architecture
The system is backed by a PostgreSQL database with Row Level Security enforcing tenant isolation. Core tables include applications (storing the extracted JD data), resumes (for the fit analysis baseline), interview_stages (tracking pipeline progression), and ai_usage_tracking (for quota management).
Technologies
Next.jsSupabasePostgreSQLGemini APIVercel Cron
Engineering Decisions
01
Resilient AI calls
rather than letting a single Gemini outage break the extraction flow, the app maintains an ordered fallback list of models, persists temporary blocks to the database (shared across concurrent requests), and returns clean, user-facing error messages instead of raw provider errors.
02
Parallel AI execution
extraction and resume matching don't depend on each other, so both requests fire concurrently and resolve to the same model where possible — cutting perceived wait time roughly in half.
03
Defense-in-depth data scoping
Row Level Security (RLS) is enforced strictly at the database layer (scoping all access to auth.uid()), while server-side API routes explicitly re-scope user queries as a secondary defense layer against accidental cross-tenant data leaks.