job search control plane

This project is a working hiring-operations system: it finds software roles, scores fit with an LLM, tracks the application funnel, and generates per-job tailored LaTeX resumes from a single source template.

email me for passwordaccess demo

The live dashboard is password protected because it contains active job data, notes, scoring rationale, and resume material. Request the password by email before opening the private control plane.

Python daemon

scheduled search ingestion and scoring loop

Postgres

single source of truth for results, statuses, and resume versions

FastAPI

authenticated JSON API with one transaction per request

Next.js

cyberpunk control plane for review, tracking, and resume generation

Claude

job fit analysis and resume tailoring

collect

A Python daemon queries multiple job sources, deduplicates the feed, and stores each result in Postgres.

score

Claude reviews each role against target criteria, writes a score, and keeps the reasoning visible for audit.

track

FastAPI exposes a private control plane for saved roles, applications, notes, status changes, and history.

tailor

The resume builder uses the saved LaTeX template plus the job context to generate a targeted resume version.

motivation

Job searching is repetitive operational work. I built this to turn that process into a reliable system: collect the noisy feed, rank the roles worth attention, preserve context, and reduce the manual work needed to respond thoughtfully.

The project also demonstrates how I approach product engineering: define the workflow, keep the data model simple, automate the expensive judgment calls, and build an interface that makes the next action obvious.

architecture signals