"Companies use AI to filter candidates. I just gave candidates AI to choose companies." —— Santiago Fernández de Valderrama
This is the ** #95 article** in our "One Open Source Project a Day" series. Today, we dive deep into Career-Ops (career-ops.org).
If 2024 was the year of the AI Agent, 2025 and beyond is the era where AI Agents penetrate every vertical. Career-Ops is a prime example, automating the entire job search lifecycle: searching, screening, tailoring, applying, and tracking. The creator, Santiago, used this system to evaluate 740+ listings and generate 100+ tailored CVs, eventually landing a Head of Applied AI role.
What You Will Learn
- Job Workflow Automation: How to leverage AI Agents to handle massive job volumes.
- Agentic Evaluation: Moving beyond keyword matching to deep semantic alignment.
- Dynamic PDF Generation: Using Playwright for real-time, ATS-optimized document rendering.
- Terminal UI Dashboards: Elegantly managing complex pipelines using Go.
- Multi-Model Integration: Best practices for integrating Claude Code, Gemini CLI, and OpenCode.
Prerequisites
- Basic understanding of AI concepts (LLMs, Agents).
- Node.js environment (for local development).
- Familiarity with CI/CD and basic CLI operations.
Project Background
Overview
Career-Ops (or careerops) turns any AI coding CLI into a full-scale recruitment command center. Instead of messy Excel sheets, you get an AI-driven pipeline from discovery to interview prep.
Problem Solved
- The Noise: Manually filtering hundreds of daily listings is exhausting.
- The Generic Trap: One-size-fits-all resumes fail modern ATS filters.
- Tracking Chaos: Keeping tabs on dozens of applications is a full-time job.
Author & Stats
- Author: Santiago Fernández de Valderrama (@santifer), Head of Applied AI.
- GitHub Stars: 53.6k+
- Forks: 10.7k+
- License: MIT
Key Features
Core Philosophy
Career-Ops is a filter, not a spammer. It emphasizes quality over quantity, using a 10-point weighted A-F scoring system to identify roles actually worth your time.
Main Modules
- The Scout (Scanner): Automatically scans 45+ top-tier portals (Anthropic, OpenAI, Mistral) and major ATS like Greenhouse/Lever.
-
The Evaluator: Extracts JD data, compares it against your
cv.md, identifies gaps, and suggests STAR-method interview prep. - The Tailor (PDF Gen): Uses Playwright to render keyword-optimized, ATS-friendly resumes in real-time.
- The Tower (Dashboard): A high-performance Go-based TUI to view, filter, and manage your pipeline.
Quick Start
# Fastest way to install
npx @santifer/career-ops init
# Enter directory
cd career-ops
# Start your preferred AI CLI
claude
# Run your first evaluation
/career-ops "https://jobs.lever.co/company/role-id"
Technical Deep Dive
Architecture
Career-Ops follows an "Agent-as-a-Framework" design. It doesn't run a traditional backend; instead, it lives as a set of structured modes and skills injected into tools like Claude Code.
Data Flow
- URL/Text Input -> Classifier
- Archetype Detection (LLMOps/PM/SDE)
- A-F Evaluation (Semantic comparison with your CV)
-
Artifact Generation:
.mdReport,.pdfCV,.tsvTracker Sync.
Notable Implementation Details
- Playwright-based Scraping: Bypasses anti-bot measures on modern career portals.
- Markdown-first Data: Uses human-readable Markdown and TSV for tracking, making it developer-friendly and version-controllable.
- Charmbracelet TUI: Built with Bubble Tea for a "Catppuccin" themed, high-refresh terminal experience.
Resources
- 🌟 GitHub: santifer/career-ops
- 📚 Documentation: career-ops.org
- 🎥 Demo: Watch GIF
- 📝 Case Study: How the author landed his role
Summary
Career-Ops shifts the power dynamic in recruitment. For the first time, candidates have a tool as powerful as (or more powerful than) the HR systems filtering them. It’s an essential project for anyone looking to master AI Agent workflows or optimize their next career move.
- GitHub Repository: santifer/career-ops
- Website: career-ops.org
Explore PrimeSkills —— A curated marketplace for AI Agents and skills, verified in real enterprise workflows. No fluff, just what works.
Visit my Homepage for more insights and interesting products.
The "One Open Source Project a Day" series - helping developers stay ahead of the curve.
Top comments (0)