@xameer Neat! I don't suppose you have an job.opml you can share?
I need to do more building of personalised tools like this myself - great work!
@jonocarroll I can or may be we need make a collective resource for everyone where people share it for everyone .
You can raise and issue and submit a pr
and why just opml , there could another api like greenhouse
The resource can have an id
With 2 sets of metadata
A- set of skills and profile
B - target roles , Companies , location 's etc
C- opml , or api
Have a look at score.rs
As of now its just sigmoid ( logistic regression) , with diverse date , I can train that with collaborative filtering or some other algo
Or do share the tool when you build one
@jonocarroll yes and its very easy to tailor an ompl file with n RSS feed URLs , even when n is one.
I intentionally didn't commit ompl , as that's my personal handcurate data , for my personal fit.
But yes , we can make a community fit or similar params.
So long as statements re truthful.
Even by #gdpr
When a resume is shared (e.g., by a recruiter to a company), it becomes personal data controlled and processed by third parties. GDPR aligns with "good faith" and penalizes "falsification" through several principles:
Accuracy (Article 5(1)(d)): GDPR requires that personal data must be accurate and kept up to date. If a candidate provides false information, they violate the requirement for accurate data.
Fairness and Lawfulness (Article 5(1)(a)): Data must be processed fairly. A recruiter sharing a falsified resume in bad faith breaches the obligation of fairness to the prospective employer.
Purpose Limitation (Article 5(1)(b)): Resumes must only be used for the specific purpose of the job application. Falsifying a resume to fit a different, unrelated role breaches this limitation.
In a strict mathematical or analytical sense, a resume acts primarily as an epistemological document, although it relies on ontological assumptions regarding the existence of the applicant's skills and history.
another release
CLI -> dsl as of now it is just tiny, typed, functional WHERE clause
1. Query languages
#SQL (but simpler, no joins)
#Lucene / Elasticsearch query DSL
#jq (JSON filtering)
#GitHub search syntax
2. Predicate DSLs
#Kubernetes label selectors
#Terraform/HCL expressions
Bazel/Starlark (restricted Python)
Like I said I ll blog about it
1. why I am building it
https://xameer.gitlab.io/post/2026-05-03-Causal-Authorship-and-Verification-Ceiling.html
2. some dev notes of how it works and what next ( my favorite and long lost #semantics and kinda sorta #typetheory
https://xameer.gitlab.io/post/2026-04-22-A-DSL-for-RSS-Feeds.html
HoldMyType
Jonathan Carroll
Antonio Patriarca