Changing careers? A CV hides what you have. A graph shows it.
A CV is ordered by employer, so the skills that would get you into the new field are scattered under job titles that do not match it. Reordering the same facts by what you know changes what a company sees.
3 min
The hardest part of changing fields is not learning the new thing. It is that the document you are judged on is organized in the worst possible way for your case.
A CV is ordered by employer. Three jobs, three blocks, each headed by a title. If you are moving from support engineering into data, every piece of evidence that you can do the new job is scattered across blocks headed by the old one — and a reader spends four seconds deciding whether the headings match the role. They do not. You are out.
Reorder the same facts by what you know, and the picture changes completely. Nothing new is claimed. The same work, grouped differently.
What that looks like
Take a support engineer moving toward data engineering. The CV says:
Support Engineer, 2021–2026 — Handled escalations for a payments platform. Worked with the engineering team on recurring issues.
Two sentences, one title, nothing a data team would stop on.
Now the same five years as claims, each with the sentence it came from:
| What | Since | The sentence behind it |
|---|---|---|
| SQL | 2021 | "I wrote the queries we used to find duplicate charges — eventually a view the whole team used." |
| Python | 2022 | "I automated the weekly reconciliation report; it used to take me a day." |
| Data quality | 2022 | "Most escalations were upstream data problems, so I built the checks that caught them before customers did." |
| Postgres | 2023 | "When billing moved to Postgres I did the query rewrites." |
| Incident response | 2021 | "I was on the escalation rota for four years." |
Every row was already true. None of it was visible. The second table is what a data team reads as this person has been doing a piece of our job for four years.
How to do it in ten minutes
Talk, do not fill in a form. Describe what you actually did — projects,
problems, the thing that used to take a day. The agent asks back where
something is unclear and never guesses a date; unknown is a correct answer and
it will ask rather than invent one.
Read what it made. Every claim shows the sentence it came from. If the extraction read led where you wrote joined, say so; the correction is recorded as a new claim with today's date, and the old one is kept, because a record with a history is worth more than one that quietly changes.
Match yourself against a real listing in the new field. Not to apply — to read the unmet list. That list is the honest gap between where you are and where you are going, and it took you four minutes to produce.
The unmet list is the plan
This is the part people are surprised by. Half the time the gap is real, and now you know the two things to learn instead of guessing at a syllabus. The other half the gap was never there — you have done the thing, your CV simply never said so, and the fix is a sentence rather than a course.
Either way you stop applying blind.
A second use, for the same four minutes: paste your own current job description and match yourself against that. The unmet requirements are the conversation to have at your next review, with evidence.
What a company sees
Not your whole graph. The comparison: which of their requirements your claims meet, the quoted sentence behind each, and the ones nobody met. Your profile is private when you create it; being visible in the pool is a separate, dated choice you can reverse at any time.
And when a claim matters, a colleague can confirm it with a sentence of their own — which is a different thing from a one-click endorsement from somebody who never saw you work.
Build your graph — free · what you get as a person · how a match is computed