About Keryn Jobs

Keryn Jobs reads public job boards that employers run on applicant-tracking systems, keeps the roles in security, AI/ML, data science and technical architecture, and sorts them into six tracks. It is a list of links. Every title opens the employer's own posting, and you apply there.

What is listed

What each card states, and where it comes from

Where it can be done from
The countries or regions named in the posting's location, title or restrictions. A place is read with what is written next to it: "Dublin, CA" is in the US, "Sydney, New South Wales" in Australia, "Berlin, DE" in Germany, and a name that belongs to several places ("Heathrow") places nothing on its own. When a Workday board lists only "3 Locations", the posting's address names one of them and the card says so: "US + 2 more locations". If no place is named, the card says "Not stated" and the role matches no region filter: unknown is not treated as worldwide.
Work mode
Fully remote, Hybrid or On-site, taken from the job board's own field, the location, or a sentence about this role. Required office days make a role Hybrid whatever its label says, and five make it On-site. Otherwise the board's own field for the role outranks a sentence elsewhere in the posting. A role offered both ways ("New York (Hybrid); United States (Remote)") shows both and appears under both filters. Company benefits, an optional hybrid schedule, and what other roles at the company do never count. If the posting does not say, no mode is shown.
Dates
Four separate facts under Details: when the employer posted it (if the board says), when Keryn added it, when its page was last checked open, and when applications close (if stated). The card shows "Added"; it is not the posting's age.
Pay
Only what the posting publishes, in its own unit (hourly pay is not converted to a yearly figure). If a role lists different ranges by location, the card shows one and marks it; Details lists each.
Level
The seniority words in the title. An architect is not shown as Staff, and "Head of" is not shown as an executive. A CISO named only as who the role supports ("CISO Focus", "Office of the CISO", "Chief of Staff to the CISO") is not an executive, and a bank's "Vice President" is shown as Senior. If the title states none, none is shown.
Employer
The company that posted the role. A recruiter posting for an unnamed client is labelled as one. Marketplaces that also hire their own staff, such as Turing and Mercor, are employers unless the posting says the role is for a client.
One role
Copies of one requisition on several career sites of the same employer are one entry, and the same title in several places is one entry with a link, place and pay for each place.
Closing
The board's own end date or the posting page's closing date, when it gives one. A role whose own closing date has passed is removed. No date is guessed.
Focus and specialty
On the AI security, research and leadership pages, filters for what the role is about: security of AI, AI for security, or AI governance; a research specialty (vulnerability and exploit, agent security, AI safety, telemetry and detection, biosecurity) and academic or product setting; and whether a leadership role runs the security function or supports a CISO. Each shows only where the posting says so.
Requirements
Degree, student level, experience, citizenship, clearance, sponsorship, languages and travel appear under Details when the posting states them, with its words. Otherwise "Not stated". A degree is read one statement at a time: "pursuing a bachelor's" says who may apply while studying, "a bachelor's degree" is one to hold already, and a degree that is preferred or "or equivalent experience" is neither. Legal notices (equal-opportunity text, the polygraph-act poster, export-licence wording) are not requirements.

Nothing here is about a person. The site holds no recruiter or hiring-manager names, no contact details and no accounts, and it sets no cookies.

The six tracks

Interns & early career

For
Students, new graduates and people in their first security or ML job.
Included
Titles that say intern, co-op, working student, apprentice, new grad, graduate, entry level, junior, residency or fellowship, in security, AI/ML or data science.
Left out
A senior word next to an early word ("Senior Associate"), manager and director titles, and AI labelling gigs.
Order
By the date the role was added, newest first, and nothing else. Roles added on the same day are ordered by pay listed, remote work and a security focus.

Research

For
Security researchers, AI safety researchers and research engineers.
Included
Titles that say research, researcher, research scientist, research engineer, postdoc or fellowship, in security or AI/ML.
Left out
User and market research, clinical and lab research, research coordinators and analysts.
Order
By how much the role is research in this field: security research and AI security in the title first, then published work, AI safety and attack research named in the description. Age matters less here, because good research roles stay open a long time.

AI security

For
People who secure AI systems or build AI into security work.
Included
Titles that pair an AI word with a security word (AI security, LLM red team, AI safety, model risk), and security roles whose description is about attacks on models: prompt injection, jailbreaks, poisoning.
Left out
Product and programme managers, and postings that use AI words with nothing behind them.
Order
By specificity: how much of the job is AI security. Security, safety and red-team wording ranks above governance wording. Postings that bundle several jobs into one rank lower.

Architecture

For
Architects: people whose job is design authority.
Included
Architect titles in four families: security, AI/ML, cloud and platform, enterprise and solutions.
Left out
Client, customer, partner, pre-sales and professional-services architects; vendor-package architects (Salesforce, SAP, ServiceNow and similar); junior titles. Solutions and consulting architects are listed, labelled as such, and ranked below in-house design roles.
Order
By family, in that order, then by evidence that design is the work: reference architectures, design reviews, threat modelling, zero trust.

ML & data science

For
ML engineers and data scientists who build and ship models.
Included
ML engineer, data scientist, applied scientist, research scientist, MLOps, statistician and quantitative researcher titles.
Left out
Analyst titles, labelling and rating gigs, product managers, security engineers who work on AI, and postings whose text shows no modelling work (no training, evaluation or ML tooling).
Order
By evidence of real modelling work: training, experimentation and named ML tooling in the description. Software roles that only sit on an AI team rank lower.

Leadership

For
People who run security, AI/ML and data science teams.
Included
CISO, chief, VP, head of, director and team-manager titles in security, privacy, trust and safety, AI/ML, data science and research.
Left out
Product, programme and project managers; sales, partnership and marketing directors; field and advisory CISOs, who speak for a vendor to its customers.
Order
By scope: C-level and VP, then head of, then director, then manager. Postings that describe reports, hiring and headcount rank higher.

Feeds

Each track has an RSS feed and a JSON feed with its 100 most recently added roles. The full list of every track is also published as JSON, with the same fields the cards show.

TrackRSSJSON FeedFull list (JSON)
All tracksfeed.xmlfeed.jsonindex.json
Interns & early careerfeed.xmlfeed.jsonearly-career.json
Researchfeed.xmlfeed.jsonresearch.json
AI securityfeed.xmlfeed.jsonai-security.json
Architecturefeed.xmlfeed.jsonarchitecture.json
ML & data sciencefeed.xmlfeed.jsonml-data-science.json
Leadershipfeed.xmlfeed.jsonleadership.json

Corrections

Listings follow the employer's board. When an employer changes or removes a posting, the entry here changes on the next rebuild and drops off within four days of the posting disappearing. To report a security problem with this site, see keryn.dev/security. This page was built 2026-11-22 14:28 UTC.