Structured evaluation
Scores, dimensions, and a crisp apply/maybe/skip signal—with rationale you can sanity-check before you invest time.
Open source · runs on your hardware
Your Job Search Orchestrated Locally
Tailor resumes, evaluate roles, and run a real pipeline—without handing your career data to a SaaS middleman. Bring your own LLM keys; the stack stays yours.
View on GitHub See why JobsHunt
MIT licensed
Fork, audit, extend
Local by default
YAML + files you own
FastAPI + React
One CLI, one UI bundle
Build from source
From a terminal, clone the repo and run the installer for your OS (Python 3.9+; Node 18+ only if the UI must be rebuilt). Full details in the README.
macOS & Linux
Same steps on both: from the repo root (folder with pyproject.toml), zsh or bash. Use
python3 as your distro provides.
zsh · bash
git clone https://github.com/ramakrishnanayyappan/jobshunt.git cd jobshunt ./scripts/install.sh
Windows
PowerShell from repo root. If execution policy blocks activation, use cmd and
.venv\Scripts\activate.bat.
PowerShell
git clone https://github.com/ramakrishnanayyappan/jobshunt.git cd jobshunt .\scripts\install.ps1
Why JobsHunt?
Local-first control and honest trade-offs next to typical hosted job-AI tools.
| JobsHunt | Typical SaaS job AI | |
|---|---|---|
| Data residency | Files and YAML on your machine; you choose paths. | Often cloud-stored posting history and drafts. |
| API keys / models | Bring your own providers; no vendor lock-in for models. | Usually bundled or opaque model routing. |
| Open source | MIT — inspect, fork, extend. | Closed code and product roadmap. |
| Subscription | Free as in software; you pay APIs you enable. | Recurring product fees common. |
| Works offline | UI and file workflows work offline; LLM features need network. | Typically requires connectivity. |
| Export ownership | Artifacts on disk; export TXT/PDF/DOCX locally. | Export and portability vary by vendor. |
Everything you need for a serious search
Built for people who want control: one workspace, clear artifacts, optional AI where you plug in providers.
Drafting & refine
Tailored drafts from your vault, optional ATS-style refine passes when you want tighter prose.
Insights & copilot
Workspace-aware suggestions and chat—confirm before anything touches your draft.
Exports
TXT always; PDF/DOCX when you install the export extras—artifacts land where you expect.
Pipeline
Track stages, links, and notes per workspace—so “where did I apply?” has one answer.
Story bank
STAR-style pins that can feed drafts—reuse your best proof without copy-paste chaos.
Batch
Queue multiple postings and let the worker chew through them—bounded, predictable runs.
Scout (optional)
Experimental link discovery—opt-in, respect sites’ rules, verify anything important yourself.
Workspaces
Split searches: different vaults, pipelines, and batches without mixing contexts.
Outreach
Template-aware personalization—copy from the UI; you stay in control of what gets sent.
How it works
Three beats—from materials to motion—without a hosted account.
01
Point at your vault
Files or folders; optional vault summary for leaner context when models charge by the token.
02
Capture the role
URL or paste—keep the spec, quirks, and must-haves beside the workspace it belongs to.
03
Draft, score, ship
Iterate on drafts, read evaluations, export, and advance rows in the pipeline you define.
Shape the project
Issues, docs PRs, and rough edges welcome. JobsHunt is built to be hacked on from your own machine.
Say hello on GitHub
No form, no tracking—just the repo, README, and issues.