Eleduck Jobs Scraper — China Remote Jobs

Get structured remote-job data out of China's biggest remote-work community.

Eleduck (电鸭社区) is where Chinese developers, designers and product people actually look for remote work. Its jobs board holds 11,000+ listings across 447 pages — real salaries, real companies, updated daily. There's no public export, no RSS, and the jobs page itself sits behind a CAPTCHA.

This Actor turns that board into clean JSON.

{
  "title": "Remote（AI agent方向）-AI工程师、后端工程师、开发者关系工程师",
  "url": "https://eleduck.com/posts/XNfwom",
  "work_arrangement": ["全职远程"],
  "job_function": ["AI工程师"],
  "hiring_type": ["企业直招"],
  "industry": ["AI/人工智能"],
  "salary_min": 20000,
  "salary_max": 60000,
  "salary_currency": "CNY",
  "salary_period": "month",
  "salary_min_cny_month": 20000,
  "salary_max_cny_month": 60000,
  "salary_raw": "20k-60k",
  "published_at": "2026-07-06T14:10:28+08:00",
  "views_count": 888,
  "comments_count": 21,
  "matched_filter": "全职远程 + AI工程师",
  "content_markdown": "## 公司介绍\n\nAIsa 成立于 2025 年..."
}

41 fields per job. Full description in Markdown and HTML. Ready to load into a database, a spreadsheet, or an LLM prompt.

---

Three things that make this different

1. Multi-select filters — something the website itself can't do

Eleduck lets you pick one value per filter. Want AI engineers and developers? Full-time remote and part-time? You're running searches by hand and merging them yourself.

This Actor takes your selections, runs every combination, and hands back one deduplicated dataset:

{
  "jobTypes":   ["19", "18"],
  "skillTypes": ["162", "10"]
}

→ 全职远程 + AI工程师      31 jobs
→ 全职远程 + 开发       5,089 jobs
→ 线上兼职 + AI工程师      34 jobs
→ 线上兼职 + 开发       2,927 jobs
   deduplicated into a single clean output

Every record carries matched_filter, so you always know which combination surfaced it. Five filter groups, 48 options total, all multi-select:

Group	Options
Work arrangement	Full-time remote · On-site · Part-time online · Part-time on-site · Hybrid
Job function	AI engineer · Developer · Product · Design · Ops · Writing · DevOps · Other
Hiring type	Direct employer · Outsourced gig · Referral · Co-founder · Recruiter
City	Overseas · Beijing · Shanghai · Shenzhen · Hangzhou … (18)
Industry	AI · SaaS · Open source · Media · E-commerce · Web3 … (12)

2. Salaries you can actually sort by

Chinese job posts write pay however they feel like it. This Actor parses it anyway:

薪资35-42K          →  35,000 – 42,000 CNY/month
18K - 35K / 月      →  18,000 – 35,000 CNY/month
年薪20W-30W         → 200,000 – 300,000 CNY/year  → 16,667–25,000/month
500～1500 美元/月    →     500 –   1,500 USD/month →  3,600–10,800 CNY
1.5万-2.5万         →  15,000 – 25,000 CNY/month
预算300–600/月       →     300 –     600 CNY/month

Half-width and full-width dashes, ～, thousands separators, K/千/W/万, five currencies, four pay periods — all normalised into salary_min_cny_month / salary_max_cny_month. One column you can sort across every format.

And it refuses to guess. These all return no salary rather than a wrong number:

2-3个真实实操案例        项目周期：约 2 - 3 周      从 0 到 1 做一个长期项目
70%基础薪资              1 - 3 年经验              填 10 分钟问卷

Every false positive above was found by checking output against live data by hand, then locked down with a test. 34 unit tests, all passing. Roughly half of real posts yield a parseable salary — the rest genuinely say "面议" (negotiable). has_salary tells you which, and salary_source_line shows the exact line every number came from, so you can verify any of it yourself.

3. No browser, no CAPTCHA, no proxies

The jobs page redirects humans to an Aliyun CAPTCHA. This Actor doesn't touch it — it reads the JSON API directly. No login, no cookies, no headless Chrome.

That means ~13 seconds and 0.014 compute units for a typical run, and nothing to break when the site restyles its frontend.

---

Works out of the box

Hit Start and you get recent AI/automation jobs immediately. Defaults are pre-filled with AI Agent, RAG, 邮件分类, 客服系统, 自动化, 爬虫 and a rolling 60-day window — recalculated every run, so a scheduled task never drifts into stale data. Clear or change any of it.

Common setups

Daily AI-jobs digest — schedule it, get only what's new:

{
  "skillTypes": ["162"],
  "lastDays": 1,
  "fetchFullContent": true
}

Paging stops the moment it crosses your cutoff, so incremental runs cost a handful of requests.

Senior roles only, salary stated:

{
  "jobTypes": ["19"],
  "minMonthlySalaryCny": 25000,
  "requireSalary": true
}

Everything, no filters — all 11,000+ jobs:

{ "maxItems": 0, "lastDays": 0 }

---

Input

Option	Default	What it does
board	Jobs Channel	Jobs Channel (11,000+ structured listings) or Community posts (20,000+)
jobTypes	—	Work arrangement. Multi-select
skillTypes	—	Job function. Multi-select
hireTypes	—	Hiring type. Multi-select
cities	—	Location. Multi-select
fields	—	Industry. Multi-select
maxItems	100	Cap results. 0 = no limit
fetchFullContent	true	Fetch full description (1 request per job). Off = 25 jobs per request, metadata only
lastDays	60	Rolling window in days. 0 = disable
publishedAfter	—	Absolute cutoff (YYYY-MM-DD) if you need one
keywords	pre-filled	Keep posts matching ANY keyword
minMonthlySalaryCny	0	Minimum monthly pay in CNY, currencies normalised first
requireSalary	false	Drop posts with no stated salary
excludeClosed	true	Skip filled positions
sort	Newest	Newest or Hottest

Advanced settings cover request delay, concurrency and proxy. The defaults are tested — you shouldn't need them.

Output

41 fields per job:

Identity — id · url · title · published_at · modified_at Structured tags — work_arrangement · job_function · hiring_type · city · industry · all_tags Salary — salary_min · salary_max · salary_currency · salary_period · salary_min_cny_month · salary_max_cny_month · salary_raw · salary_source_line · has_salary Content — content_markdown · content_html · content_text · summary Signals — views_count · comments_count · upvotes_count · is_closed · is_featured · is_pinned Poster — author_nickname · author_level · author_roles Provenance — matched_filter · scraped_at

Also available as CSV, Excel, XML or RSS through the Apify API — and every run exposes a runSummary reporting what each filter combination matched and why anything was skipped.

Cost

Pure HTTP, so runs are cheap:

Mode	Requests
Metadata only	1 per 25 jobs
With full descriptions	1 per job
Daily incremental (lastDays: 1)	a handful

A 25-job run with full text: ~13 seconds, 0.014 CU.

Built for

• Job boards & aggregators — syndicate China remote listings
• Recruiters — track who's hiring, at what pay, in your niche
• AI agents — output schema is defined, so agents can chain this Actor via MCP or API without guessing the shape
• Market researchers — salary benchmarks and hiring-demand trends by role and industry
• Job seekers — a filtered feed instead of refreshing a page

Notes

Public data only. No emails, phone numbers or contact details are collected, in line with Apify's personal-data policy. Public nicknames and engagement counts are included.

Only English-language field names are used; job content stays in its original language (mostly Chinese, some English).

Questions or a field you need? Open an issue on the Actor page.