How to Write a Resume in 2026: The Modern ATS + AI Guide (with examples)

What actually changed in 2026

Three things shifted in the last two years and most candidates haven't caught up. First, ATS parsing got stricter, not smarter — the old `Word art header, two-column theme, embedded icon` resume that used to get through is now scored as malformed text. Second, a second layer of LLM screening sits on top of ATS at most companies hiring more than 50 people a year: a rubric prompt compares your resume to the job description and assigns a 0–100 score. Third, recruiters spend less time reading — the average first-pass scan dropped from 7.4 seconds (2018, TheLadders) to under 5 seconds in 2025 industry surveys. You have less time than ever.

The fix isn't to make your resume fancier. It's to make it boringly correct: single-column, clean text, well-named sections, and ruthlessly specific bullets that read well to both software and a tired human at 9 PM on a Friday.

The structure that works in 2026

Use this exact section order unless you have a strong, specific reason not to. It mirrors how ATS parsers expect to find data and how recruiters scan.

1. Header: Name, one-line title, city/country, email, phone, LinkedIn, portfolio. No photo (illegal to consider in most countries, ignored everywhere else).
2. Summary (3 lines max): role + years + your strongest specific outcome. Skip the “motivated team player” fluff.
3. Experience: newest first. Company → role → dates → 3–5 bullets. Each bullet has a verb, a number, and a noun.
4. Projects (optional but powerful): one to three side projects, especially for early-career.
5. Education: degree, institution, year. Add coursework only if you graduated <3 years ago.
6. Skills: a clean comma-separated list of the actual technologies, tools, and frameworks. No skill bars, no star ratings.
7. Optional: Certifications, Publications, Awards, Languages.

How to write a bullet recruiters remember

Every weak resume bullet I've ever seen breaks one of three rules. Every strong one follows all three: verb → number → noun.

Weak: “Worked on the dashboard team to improve the user experience.”

Stronger: “Led the rewrite of the analytics dashboard (React + Tanstack Query), cutting median load time 47% and growing weekly active users 2.3× across 80k accounts.”

Notice what changed: a precise verb (Led), specific technology, two numbers, and a scoped audience. If a bullet doesn't have a number, ask yourself: how would I prove this in an interview? Whatever the answer is, put a version of it in the bullet.

Formatting rules ATS parsers care about

• Single column. Two-column layouts get re-flowed in ways you can't predict. Date columns become attached to the wrong job.
• No tables, text boxes, or embedded images. ATS parsers extract plain text. Anything else is risk.
• Standard section names. Use “Experience” not “Where I've made a dent.” Cute headers get skipped by the parser.
• PDF, not Word. But export from a tool that produces text-selectable PDFs — not a flattened image of one.
• Standard fonts. Inter, Roboto, Calibri, Arial, Times New Roman, Garamond. Anything else risks substitution.
• MM/YYYY date format. “Jan 2024 – Present” parses better than “01/24 – Now.”

Keywords without keyword stuffing

The lazy advice is “put the job description's keywords in your resume.” The right advice is “put the job description's keywords in your resume where they're true.” ATS scoring rewards keyword density, but a human recruiter reading a resume packed with terms you've never used in context will throw it out at first interview.

The workflow we recommend: paste the job description into a tool like CraftMyResume's ATS Match scorer, get the list of high-frequency terms it expects, then revise individual bullets to use the terminology you actually did. If the JD says “gRPC” and you wrote “our internal RPC framework,” change it. If the JD says “OKRs” and you've genuinely run OKRs, say so explicitly.

Where AI helps (and where it doesn't)

After helping thousands of users edit resumes with AI, here's the honest split:

AI is great at:

• Rewriting a vague bullet you've drafted into something tighter and more specific.
• Mining a job description for the keyword set the ATS will look for.
• Generating a first-draft cover letter that ties your wins to a JD — which you then heavily edit.
• Flagging passive voice, weak verbs, and inconsistent tense.

AI is bad at:

• Inventing achievements. Every AI-fabricated number is a landmine in your next interview.
• Designing the layout. Ask AI for “a beautiful resume template,” get an ATS-hostile masterpiece every time.
• Knowing what's actually impressive in your domain. A senior recruiter for SRE roles will smell a generic “owned reliability” bullet in three seconds.

The eight mistakes we still see every week

1. An objective statement. Replaced by the summary in 2010. Delete.
2. “References available upon request.” Implied. Delete.
3. Personal data — date of birth, marital status, religion, photo. Not just useless, often actively scrubbed by EEO-compliant ATS configurations.
4. Skill bars and star ratings. You're claiming a precise quantitative measurement of yourself. Just list the skills.
5. Color blocks behind text. Parsers sometimes drop the text inside them entirely.
6. Headers and footers. Many ATS skip everything inside them. Put your contact info in the body.
7. “Hard worker, team player, fast learner.” Every candidate claims these. Show them in a bullet instead.
8. Lying. Inflated titles and invented metrics get caught at the back-channel reference and end careers. Don't.

Where to go from here

If you want to skip the formatting trap entirely, CraftMyResume gives you 15+ ATS-friendly templates, a side-by-side ATS Match scorer against any JD, AI bullet rewrites that don't invent numbers, and an application tracker so you never lose a thread. It's free to start — 20 credits, no card.