Job search in the AI era

The 2026 tech CV: a structure that passes both machines and humans

Published: 2026-07-09

Nir Kosover · Founder of JobFox. Building the job-search engine for Israeli tech.

Most CV advice is about what to write. Far less of it is about how to arrange that on the page, which is exactly the part that decides both whether a machine can read you and whether a human wants to. The same file passes through two very different readers, and good structure is what satisfies both without compromising for either. Here is what that looks like in practice, including the local norms worth knowing if you are job hunting in Israel.

Why one file must satisfy both a parser and a human

Because both read it, one after the other, and each needs something different. The parser reads first: it pulls the text out of your file and sorts it into fields. Only after it succeeds does a human recruiter scan whatever is left, with tired eyes, in about ten seconds. If the structure fails the first reader, the second never sees you, and if it passes but confuses the second, that costs you too.

The good news: there are not two conflicting structures here. A clean single-column layout, with clear hierarchy and orderly dates, is exactly what the parser needs and also what helps a human find the information fast. How the ATS pipeline itself works, step by step, is covered in the companion piece, What Really Happens to Your CV in an ATS in Israel. This article is about the structure of the file itself.

The structure that works: six principles

Each principle has a technical reason (why the machine needs it) and a human reason (why it also helps the recruiter). This is not a list of compromises, it is six places where the two needs line up:

1. Single column. Jobscan recommends a top-to-bottom, single-column layout, because most parsers read left to right across the page, and a two-column design gets read as scrambled text. For a human, one column is also easier to scan quickly without jumping back and forth between sides.
2. Real selectable text, not an image and not table-heavy. A parser can only extract live text. A file exported from a design tool as an image layer, or one where all the content is crammed into tables, loses most of its data. For a human, live text also means they can search the document and copy a contact detail.
3. Standard section headings the parser expects. “Work Experience”, “Education”, “Skills”, not “My Journey”. The parser looks for the familiar words to know where each part begins. A human recruiter scans by those same headings and expects to find them in the usual place.
4. Reverse-chronological with clear dates. Most recent role first, each role with a start and end date in a consistent format (Month Year, and “Present” for the current role). The system reads dates to calculate tenure, and it guesses wrong when they are ambiguous. The human, for their part, wants to see at a glance what you have done lately.
5. A plain contact line in the body. Name, phone, email, and LinkedIn on a normal line at the top of the page, not inside a header or footer. Many parsers skip the header and footer region, and then your phone and email simply vanish. For a human, one clear line saves a search.
6. Consistent formatting. The same font, the same sizes for headings and body, the same bullet style throughout. Consistency helps the parser recognize a repeating pattern, and to a human it signals order and attention to detail without them noticing why.

One small principle earns all the rest: the job title. A Jobscan analysis of over 2.5 million applications found that CVs including the exact job title from the posting received 10.6 times more interview invitations. If the role is called “Backend Engineer”, write exactly that, not “server developer”.

What to include and cut

Now for the content, and especially the local norms that trip up people who have seen templates from abroad. Here is what belongs in and what you can quietly leave out:

1. Length: one to two pages. Focused beats full. A third page almost always dilutes what matters instead of adding to it.
2. Photo: not expected, safe to drop. In Israeli tech a photo is optional, and many candidates skip it. For the ATS version, leave it out, since the parser ignores it anyway.
3. Military or national service: short and plain. A line with the unit or role and the dates. If it is technically relevant to the job, one achievement line. No personal story.
4. Skip irrelevant personal details. ID number, marital status, age, and full home address add nothing and sometimes hurt. A general city of residence is enough.
5. Results over responsibilities. Instead of “responsible for the data pipeline”, write what changed because of you: “cut processing time by 40%”. A concrete number convinces both the machine (specific language) and the human (proof, not a claim).

How to check it actually works

Do not guess. Three simple tests show you what the machine and the human actually pick up:

1. The paste-into-plain-text test. Open the PDF, select all, copy, and paste into a blank text document. If the order scrambles, words run together, or your contact details disappear, that is exactly what the parser receives. If the text comes out clean and in the right order, you have passed the critical step.
2. The parse preview. Some systems, Greenhouse chief among them, show the candidate the fields the system extracted before you submit. If an employer offers it, do not skip it: it is your chance to catch a field that parsed wrong.
3. Send it to one human and ask what they remember. Give the file to someone who does not know you, let them look for half a minute, then ask what stuck. If they cannot recall your last role or your headline achievement, the structure is burying the most important things, which is a fix of order and emphasis, not of content.

The bottom line: good structure is not a compromise between machine and human, it is where they meet. Single column, live text, familiar headings, clear dates, and results over responsibilities, and you have already passed both readers. JobFox’s CV tools build the file in a single column and show you which fields an ATS extracts from it, so you know the structure works before you submit.