The Vision: Free AI-Powered Job Search for Everyone

[santifer]

The problem

Job searching is broken. Companies use AI to filter you out before a human ever sees your resume. Meanwhile, tools that help you fight back charge $20–140/month — to people who are, by definition, looking for income.

58% of professionals say finding a job has gotten significantly harder. Applications per role have doubled since 2022. And yet the "solutions" are paywalled.

That's not right.

What we're building

career-ops started as my personal system — I used it to evaluate 740+ job offers, generate 100+ tailored CVs, and land my current role. I open-sourced it because I believe finding a job should be free.

Today it's a powerful CLI tool for technical users. But the vision is bigger:

career-ops will become a free, local-first AI job search agent that anyone can run on their own computer — no API keys, no subscriptions, no data leaving your machine.

Roadmap

🟢 Now (what we're shipping)

• Community foundations — Code of Conduct, Governance, Contributor Ladder, Security policy
• Zero-token portal scanner (scan.mjs) — scans 73+ company APIs with zero LLM cost
• Language expansion — Japanese, Korean, Brazilian Portuguese modes live. German and French already shipping
• Security hardening across all scripts
• Issue templates and PR templates to make contributing easier

🔵 Next (what's coming)

• Free local model support — Ollama + LM Studio integration so you can run career-ops with Llama 3, Mistral, or Phi-4 on your own machine. Zero API cost, total privacy
• Claude Desktop integration — one-click install for semi-technical users
• Setup wizard — replace YAML config files with a simple guided flow. Paste your CV, pick your target roles, done
• CLI-agnostic engine — make the evaluation engine work with any AI backend via a simple env var

🟣 Later (the big vision)

• Desktop app — a native application (Tauri-based, not Electron) that anyone can download and use. No terminal required. Drag-and-drop your CV, paste a job URL, get a full evaluation + tailored PDF in 30 seconds
• Bundled local AI — the app ships with a model built-in. Download, open, use. That's it
• Every language, every market — localized modes for India, MENA, Southeast Asia, Latin America, Eastern Europe
• Personalized learning paths — when the system detects skill gaps, it helps you close them

Why this matters

Most AI job search tools charge $20–140/month. career-ops is free, open source, and runs on your own machine. There's historical precedent for what happens when a free tool matches a paid one: the free tool wins because the community is stronger than any company.

We're going to do the same for job searching.

How you can help

You don't need to write code to contribute. Check our Contributor Ladder — every role matters:

• Translate — add your language so people in your country can use it
• Test — try career-ops and report what's confusing or broken
• Build — help with local model integration, the desktop app, or the setup wizard
• Share — if career-ops helped you land an interview or a job, tell people about it

Read CONTRIBUTING.md to get started.

One more thing

I built this because I was tired of the job search grind. This is MIT-licensed, open source, and free — built by someone who's been exactly where you are.

Let's make job searching suck less — together.

— Santiago

[aa29zzjj]

Really cool idea — the agentic job search pipeline makes a lot of sense.

I managed to get Claude Code running, but a few things were a bit rough:

• Claude Code dependency (and subscription requirement) isn't obvious at first
• /career-ops isn't a standard CLI command, which was confusing initially
• Parsing job URLs seems fragile (LinkedIn search pages break the flow)

The concept is solid though — especially the CV tailoring + evaluation loop.

Would be great to see:

• Better onboarding / error handling
• Support for non-Claude backends

Thanks for open-sourcing this!

[santifer]

Thanks for your feedback! We are working on this; better onboarding and error handling are a must.

Career-ops actually supports non-Claude backend, which one are you missing?

[Deepakhacker-lab]

Great contribution

[santifer]

Thanks!

[KrAsH-CoD3]

Great work I must say.

[TheKelvinPerez]

This is a really cool project i'm going to give it a spin an see how it works

[shvetsov-as]

100% cool project and great work!

[itsmedhawal]

Hi @santifer :)
Incredible vision laid out here. The "free, local-first AI job search agent" direction is exactly where this needs to go.

I have been building a browser-based companion to career-ops-india (my India fork of career-ops) and wanted to share what I shipped this today. It directly mirrors what you are planning and might be useful to the community.

Career-Ops-India App v7.0 — itsmedhawal.github.io/career-ops-india-app

The first app in India to offer a hybrid Ghost Buster — local + cloud, mobile-first, privacy-first.

What is live today:

• ⚡ Fast Scan — instant local GLS via Gemma 2 2B running on-device via WebLLM + WebGPU. Zero API cost. Works offline after a one-time ~1.6 GB download.
• 🔍 Deep Eval — full A–G evaluation via Claude API when the role passes the GLS pre-check
• 🔒 Total privacy — JD text never leaves the device during Fast Scan. API key encrypted with PIN using AES-GCM.
• 📱 Mobile-first PWA — installable, works offline, optimised for one-thumb use on the go

The hybrid architecture:
Local AI acts as the First Responder — fast, free, offline GLS pre-check. If the role looks real, the user then spends API credits for the deep A–G evaluation and CV tailoring. This maps exactly to your "Zero token portal scanner + LLM evaluation" split.

On the Gemma 4 E2B roadmap:
Built with a model selector so the community can upgrade to Gemma 4 E2B the moment MLC compiles it for WebLLM — no code changes needed, just a model ID swap. Watching the MLC repo for that.

India-specific additions on top of career-ops:

• Ghost Likelihood Score (GLS) — original 0–100 detector across 9 India-calibrated signals (repost, Naukri source quality, fake remote, unnamed client, etc.)
• 7 India archetypes, CTC/LPA framework, GCC stage, Fake Remote detection
• Intern Mode — PPO probability, stipend benchmarks across 10 Indian cities
• Custom CV generator and Interview Prep Mode

The lineage is clear: career-ops → career-ops-india → Career-Ops-India App. Everything builds on the foundation you created. Thank you for open-sourcing it.

Would love to contribute the India module back to the main project when the time is right — especially GLS and the browser-based local AI architecture, which could serve as a reference for other regional implementations.

— Dhawal
career-ops-india · Career-Ops-India App

[itsmedhawal]

https://www.linkedin.com/posts/dhawalshrivastava_distributedintelligence-dtalks-careeropsindia-share-7448879949538545664-nDPN?utm_source=share&utm_medium=member_desktop&rcm=ACoAAATFzxkB0oXfvXocZiJc2pihpYhHVjEFims

[santifer]

Dhawal, nice work. The local AI pre-check with Gemma is a smart approach to the "free for real" problem. Running inference on-device means zero cost for the user, which is exactly the direction we laid out in the roadmap. The India-specific ghost detection with calibrated signals also adds real value for that market.

Good to see people building on career-ops and taking it to new regions. That's what open source is about.

Feel free to reach out if you want to chat more about what you're seeing in the Indian market.

[itsmedhawal]

@santifer sure thing :)

[DanielLin9406]

I completely support this idea. I recently had both of my accounts banned by Anthropic, even though I was only using Career-Ops. It's incredibly dangerous to risk vendor lock-in; a sudden ban can completely disrupt any workflow or progress that relies on a specific AI model.