Sift

AI-powered job search and application tracker. Scrape jobs, extract resume keywords with AI, match and filter listings, track your pipeline from discovery to offer — all in a native desktop app.

Built with Wails v3 (Go + Svelte). Project intelligence managed by Evo — the engineering brain for AI-first teams.

— sift.purfectlabs.com · GitHub —

Features

• AI Resume Parsing — Upload your PDF resume. Hermes (local LLM) extracts 40-60 keywords automatically.
• Smart Job Scraping — Searches Google Jobs via SerpAPI. Multi-query support with deduplication.
• Match Scoring — 4-level keyword matching engine scores every job 0-100 against your resume.
• Remote Filter — Automatic remote/hybrid/on-site classification with positive/negative keyword scoring.
• Gap Analysis — Per-job skill gap analysis comparing your resume against each listing.
• Market Analysis — Aggregated market trends across all scraped jobs.
• Pipeline Tracking — Full pipeline: New → Saved → Applied → Interviewing → Offer.
• 100% Local — SQLite database, no cloud accounts, no data leaves your machine.

Screenshots

Dashboard	Settings

Jobs	Scraper

Insights

Download

Pre-built binaries for macOS, Linux, and Windows from Releases.

Quick Start

Prerequisites

Dependency	Install	Purpose
Go 1.25+	go.dev/dl	Backend runtime
Node.js 22+	nodejs.org	Frontend build
Python 3 + pymupdf	pip install pymupdf	Resume PDF parsing
Hermes CLI	Install guide	AI keyword extraction, gap analysis, market analysis
SerpAPI key	serpapi.com (free tier: 100 searches/month)	Google Jobs scraping

Hermes setup (one-time):

pip install hermes-agent        # or: brew install hermes-agent
hermes setup                    # configure your LLM provider + API key
hermes model                    # set default model (e.g. deepseek-chat)

SerpAPI setup (one-time):

1. Sign up at serpapi.com — free tier gives 100 searches/month
2. Copy your API key from the dashboard
3. Paste it in Sift Settings → SerpAPI Key, or add to ~/.sift/.env:

   SERP_API_KEY=your_key_here
   

One-liner: ./dep-install.sh installs Go, Node, Python, pymupdf, and Wails3 CLI.

# Clone
git clone https://github.com/purfect-labs/sift.git
cd sift

# Install system dependencies
./dep-install.sh

# Install frontend deps
cd frontend && npm install && cd ..

# Run in dev mode
go install github.com/wailsapp/wails/v3/cmd/wails3@latest
wails3 dev

# Build for production
task build

Build from source (no Wails CLI)

cd frontend && npm install && npm run build && cd ..
CGO_ENABLED=1 go build -o sift .
./sift

The frontend must be built first (npm run build) so it's embedded in the Go binary.

Configuration

Paste your SerpAPI key in the Settings panel, or create ~/.sift/.env:

SERP_API_KEY=your_serpapi_key

All job data is stored in ~/.sift/jobs.db (SQLite, auto-created on first run).

Architecture

sift/
├── main.go              # Wails v3 entry point
├── service.go           # JobService — core business logic (17 API methods)
├── scraper/
│   └── engine.go        # SerpAPI client, Hermes AI, keyword matching
├── db/
│   └── jobs.go          # SQLite schema, CRUD, keyword storage
├── frontend/
│   └── src/
│       └── lib/components/
│           ├── Dashboard.svelte      # Job pipeline board
│           ├── JobCard.svelte        # Individual job display
│           ├── PositionManager.svelte # Job title queries
│           ├── ScraperPanel.svelte   # Scrape controls + logs
│           ├── Insights.svelte       # Keyword stats + analysis
│           └── Settings.svelte       # Config panel
├── build/               # Cross-platform build configs
└── .evolution/          # Evo project intelligence

How It Works

1. Upload Resume → Hermes reads PDF → extracts 40-60 keywords → stored in SQLite
2. Recommend Positions → Hermes suggests 5-8 job titles to search for
3. Scrape Jobs → SerpAPI Google Jobs search → dedup via SHA256 → compute match scores
4. Filter Remote → Keyword-based classification → Hermes keyword analysis on matches
5. Review Pipeline → Dashboard shows jobs with match scores, gap analysis, status tracking
6. Market Analysis → Hermes analyzes top 20 jobs for trends and skill gaps

Tech Stack

• Backend: Go 1.25, Wails v3 alpha
• Frontend: Svelte, Vite
• Database: SQLite (modernc.org/sqlite — pure Go, no CGO)
• AI: Hermes CLI (local LLM agent)
• Job Search: SerpAPI (Google Jobs)
• PDF: Python 3 + pymupdf

License

MIT — see LICENSE