How Roleford matches you to postings
Each morning we take the night's new engineering postings, leave out the ones that can't work for you, and rank the rest by how well you fit. This is what we read, what we leave out and how the score is made.
1Read you Skills, CV, public code and what you're looking for
2Collect postings New engineering roles, every night
3Leave out Places, levels, pay and languages that don't work for you
4Rank A fit score from 0 to 100, with the reasons
What we read about you
- The skills you list, each marked as core or as one you also know. The roles you're looking for count as core.
- Your CV or LinkedIn PDF. AI reads it once and pulls out your roles, skills, languages, location and education. It leaves out your email, phone number and street address.
- Your public GitHub. The languages in your repositories, the libraries your projects depend on, and the topics and descriptions of your repositories and the ones you've contributed to. Never private repositories.
- Other public profiles you add, such as Stack Overflow or Hugging Face, which can add skills.
- Public profiles we find through your GitHub or search engines, such as your LinkedIn headline or your website.
- What you're looking for: roles, level, work mode, places, the pay you expect and any companies you never want to see.
We show you what we find and where it came from. Anything you mark as wrong stops counting.
Verified skills. A skill is verified when your public work uses it in a real way: your code, or a model or dataset you publish on Hugging Face. One passing mention isn't enough. Skills from your CV or typed by you are stated: they count, but a verified skill counts a little more.
New postings, every night
We collect postings each night from job boards and career sites, and remove the ones that have closed. We keep engineering roles only, up to 30 days old. Your list is rebuilt once a day, after the night's postings are in, and again whenever you change your skills, your CV or what you're looking for. It holds up to 200 postings.
What we leave out
A posting leaves your list only for a clear reason:
- Skills. None of the main skills it asks for is among yours.
- Work mode. It's in a mode you didn't pick, such as remote when you only take on-site, or hybrid when you only take remote.
- Level. It's far from every level you picked, such as a staff role when you asked for junior. If you picked none, we use the level we read from your CV and code.
- Language. It needs a language you don't list. English is assumed.
- Pay. It states its pay, and even the top of its range is below what you expect.
- Place. It's outside the countries and cities you named, or, if you named none, too far from where you live.
- Your own choices. A company you hid, or a posting you've already rated or started.
Your list keeps the best 200 of what's left.
A missing level, pay or language never hides a posting. It stays and simply earns fewer points for that part.
The fit score
Every posting left gets a score from 0 to 100. Skills count most: how many of the main skills the posting asks for you have, and how many of your core skills it uses. Then how strongly you hold them (verified skills count more), place and work mode, level and pay. How new the posting is counts least.
80 and above is a Strong fit, 65 to 79 a Good fit, below 65 a Fair fit. Each match shows why: the skills you share, the main skills it asks for that you don't list, and how place, level and pay compare.
- Remote in a country you named
- Senior, as you asked
- Pay meets what you expect
Your ratings
When you rate a posting, it leaves your list and moves to your Good or Bad tab. When you rate a bad match as Pay too low, Too far / wrong place or Not my stack, that part counts a little more in your list from then on. Not this company hides that company. Ratings change only your own list.
Where AI is used, and where it isn't
- AI reads your CV once. AI and simpler text rules read each posting for its skills, level and place, and AI decides which postings are engineering work.
- The same rules leave postings out and score them for everyone. Your own ratings shift how much pay, place and skills weigh in your list.
- AI writes the drafts for your applications, which you check and send yourself.
Matching ranks postings for you. It makes no decision about you.
What we miss
- Many postings don't state a level or pay. They stay on your list with fewer points, so a few may not suit you.
- A posting that doesn't say it's remote counts as on-site.
- A remote posting limited to one state or region, such as "remote in Texas", can slip through.
- A posting that doesn't say where the job is never reaches your list.
- If we don't know which country you live in, you only see postings open worldwide. Add your country in Profile.
- Skills from a CV alone are never verified. Connecting GitHub helps if your work is public.
- If you haven't opened Roleford for 14 days, your daily list pauses until you're back.
See what fits you this morning
More questions? Read the FAQ.