linkme/README.md

linkme

Scrapes LinkedIn job listings into a local CSV. Run it daily, delete rows you've applied to.

Usage

bun fetch      # scrape and append new jobs to results/jobs.csv
bun recheck    # show jobs currently filtered out + keyword suggestions

No setup needed. Results are appended to results/jobs.csv — reruns are safe, already-seen jobs are skipped by ID.

Structure

config.ts          # search keywords and areas
fetch.ts           # scraper, HTML parser, filters
fetch-other.ts     # extra sources (datacareer.ch, stellen-basel.ch)
index.ts           # orchestrator: runs searches, writes CSV
recheck.ts         # raw-mode fetch, lists filtered-out jobs
scripts/           # one-off search scripts (temp, not in config.ts)
cv/                # polo.md (profile) + CV files
results/           # jobs.csv output
jobs/              # notes on specific job leads
plans/             # LinkedIn post drafts

Output

results/jobs.csv — one row per job:

column	description
date_found	date you ran the fetch
posted_at	date the job was posted
search	which search config matched it
title	job title
company	hiring company
link	canonical job URL
source	site the job came from

Open in any spreadsheet app. Delete rows as you apply or dismiss.

How it works

• Paginated scraping — fetches up to 6 pages per search (150 jobs) with a 400 ms delay between pages
• Card-boundary-safe parsing — each job card is isolated before extracting fields
• Language filter — drops German (m/w/d, umlauts, gender markers) and French (H/F, accents) titles
• Software-role filter — requires a relevant term in the title (software, developer, data, AI, backend, frontend, devops, platform, architect, tech lead, typescript, python, rails, node.js, ML, cloud, security, SRE, etc.)
• Seniority filter — drops junior, intern, trainee, apprentice, werkstudent, entry-level
• Global deduplication — IDs indexed from existing CSV on startup; already-seen jobs are never re-added
• 30-day recency — only jobs posted in the last month

recheck

bun recheck fetches all searches in raw mode (bypassing the software-role filter) and lists every job that would be filtered out — useful for spotting false negatives. Also prints a word-frequency table of filtered titles to help decide what to add to SOFTWARE_TITLE_RE.

Temporary searches

Write a throwaway script in scripts/ instead of editing config.ts:

bun run scripts/bern-temp.ts

Scripts in scripts/ import from ../config.ts and ../fetch.ts.

Configure searches

Edit config.ts:

export const KEYWORDS = "Software Engineer OR Data Engineer OR ...";

export const searches: SearchConfig[] = [
  {
    label: "Basel (city)",
    keywords: KEYWORDS,
    geoId: "103383283",
    location: "Basel",
    workType: ["onsite", "hybrid"],
    timePosted: "month",
  },
];

Known Swiss geoIds:

• 103383283 — Basel (city)
• 90009885 — Basel Metropolitan Area
• 107117530 — Basel-Country
• 102436504 — Zürich (city)
• 107814425 — Zürich Metropolitan Area
• 90009888 — Zürich Area

You can also paste any LinkedIn search URL directly as url: instead of building from filters.

SearchConfig also supports jobType (full-time, part-time, contract, temporary, internship) and seniority (internship, entry, associate, mid-senior, director, executive).