MyCIEL

← Back to the dashboard

What is this?

MyCIEL is a personal, working remake of CIEL (Carrières Internationales en Ligne), the French Ministry for Europe and Foreign Affairs' portal of vacancies in international organisations — which is not updated anymore. MyCIEL scrapes the career portals of the organisations directly, every two days, stores everything in a local database, and highlights the postings that match a senior ICT / cybersecurity profile (keyword match + minimum grade per organisation, e.g. AD6+ for EU institutions, P4+ for the UN system, A3+ for NATO).

The dashboard lets you search, filter by organisation, seniority, domain and status, and export to CSV. The map view plots the duty stations of the matching vacancies.

Organisations covered

This table is generated live from the database.

Organisation	Jobs tracked	Relevant
Loading…

Known gaps & limitations

• EPSO only carries vacancies that EU institutions choose to route through it — agencies that publish solely on their own portals are scraped separately (ENISA, Europol, ECCC, EUSPA, SatCen, ELA, BEREC, ERA, Cedefop).
• ITER: only the new SAP career site is scraped; the legacy portal (JavaScript-only) may list additional openings.
• Not scraped: AfDB and ITC hard-block non-browser clients; CTBTO and EUIPO run SuccessFactors' SAPUI5 app (results never reach the DOM); NATO NCIA (partially covered via nato.int), UNOPS, UNFCCC, BIS, WMO, Eurojust, EU-OSHA and ECMWF are JS-only or captcha-walled; F4E's own portal (partially on EPSO) and the non-Workable CGIAR centers. UPU, IHO, BIPM and OIV publish no machine-readable vacancy list at all.
• Locations: several portals only show the duty station on the job's own page, so about 450 stored postings have no location and cannot appear on the map (mostly UNICEF, NATO, UNESCO, EBRD, ICRC). Of those that do have one, ~98% resolve to coordinates.
• EIB, IFAD and ECHA run PeopleSoft, whose result grid only exists after JavaScript runs: those three are scraped with a headless browser and are more fragile than the other sources. ECHA's portal also keeps years of closed vacancies on the same page — only ones with a future deadline are kept.

What counts as relevant

Two stages. Postings that are not staff vacancies are never stored: internships, studentships, traineeships, apprenticeships, fellowships and postdocs, graduate programmes, consultancies and secondments (including Seconded National Experts). Of what remains, a posting is flagged relevant when its title matches the ICT / cybersecurity keyword list and it meets the minimum grade for that organisation; General Service posts and locally-recruited positions outside France are kept but never flagged.

The dashboard opens on Relevant & open. Switch the Status filter to "All jobs" to browse everything that was stored, or to Bookmarked for the ones you starred.

Closed vacancies are detected two ways: a deadline in the past, or the posting disappearing from its organisation's listing. The second matters because many portals publish no deadline and keep the job page online (with a working "Apply now" button) after the vacancy is filled. Closed jobs are hidden unless you choose "All jobs", where they appear greyed out and marked delisted.

Bookmarks (the star column) are kept in your browser's local storage, so the site needs no accounts or server - but they stay in this browser only and are not synced between devices.

Matching keywords also produce the domain tags used by the filter (IT, cybersecurity, cloud, networking, AI/ML, architecture, software, data, devops, telecom, digital). Jobs that match no keyword carry no tag, so picking a domain never returns unrelated postings. Matching is done on the title only — and on the title minus any appended organisation name, so a "Legal Officer" at ENISA is not flagged just because its employer is a cybersecurity agency.

How it works

Python scrapers (requests/BeautifulSoup, plus HTTP/2 and a headless browser where a portal needs one) → SQLite → this static dashboard. The whole thing runs serverless: a scheduled GitHub Actions job scrapes every two days, keeps the SQLite database as a release asset so it can tell new postings from old ones, and publishes these pages to Cloudflare Pages. Datacenter IPs cost almost no coverage — of 63 sources only one (Frontex) refuses them, and its vacancies show up via EPSO anyway.