Most engineering resumes that get rejected fail on parsing, not on qualifications. The applicant tracking system (ATS) reads your PDF as a stream of text, and if your layout scrambles that stream, a recruiter never sees your best work. The fix is boring and effective: single-column layout, standard section headings, real text (not images), and metrics in your bullets. Do that, run a 30-second copy-paste test, and you clear the mechanical bar that most candidates trip over.
We run mock interviews all day, and the pattern we see is that strong engineers under-invest in the resume mechanics and then blame "the algorithm" when they get ghosted. The resume rarely needs to be clever. It needs to be machine-readable and specific.
What an ATS actually does with your resume
Two things happen, and people conflate them. First, the ATS parses your file into structured fields: name, contact, work history, dates, skills. Second, a recruiter or an AI feature ranks and filters those parsed records against a job.
Parsing is where formatting matters. According to Jobscan, most ATS parsers read a resume linearly, left to right and top to bottom, so side-by-side columns and tables cause the software to read straight across the page and mash unrelated content together into an unreadable block. If your job titles land in the wrong field, or your skills get glued to your address, the ranking step is working from garbage input.
The auto-reject myth is overstated. Modern systems mostly rank and surface candidates rather than silently deleting them, and a human still reviews the top of the list at most companies. But there are hard knockout filters (work authorization, location, sometimes a required credential), and a resume that parses into nonsense can effectively drop you out of contention without anyone deciding to. Your goal is to never lose points to the machine before a person reads you.
The formatting rules that survive parsing
These are the rules we would stake a callback on. None of them are about looking impressive. They are about surviving the parser.
- Single column, top to bottom. Jobscan recommends a clean single-column flow so the software correctly identifies contact info, skills, and work history without skipping sections. Two-column "designer" templates are the single most common reason a well-qualified resume parses badly.
- Contact info in the body, not the header/footer. Some parsers ignore text in document headers and footers. Put your name, email, phone, and links on the first lines of the page body.
- Standard section headings. Use "Experience," "Education," "Skills," and "Projects." Creative labels like "Where I've Made Impact" can fail to map to the expected fields.
- No tables, text boxes, or graphics for real content. Jobscan's writers are blunt that graphics, tables, and creative fonts are a reliable way to get lost in an ATS. If information matters, it must be plain text.
- Skills as a delimited list, not a grid. Separate items with commas, bullets, or vertical bars. A skills table is exactly the structure that scrambles.
- "Umbrella" formatting for promotions. When you held multiple roles at one company, list the company name once at the top so the ATS attributes every role to the correct employer.
- Export a text-based PDF. A PDF generated from Word or Google Docs keeps a real text layer. A PDF that is secretly an exported image (or a screenshot) parses as nothing.
A before/after bullet
Formatting gets you parsed. Content gets you ranked. Vague bullets waste the space you fought to keep readable.
Before:
Responsible for working on the backend and helping improve system performance.
After:
Cut p99 API latency 40% (820ms to 490ms) by adding Redis caching and batching N+1 queries across 3 services.
The second version parses into the same fields, but it hands the recruiter (and any keyword match) concrete verbs, numbers, and technologies.
The 60-second parse test
Do not trust how your resume looks. Test how it reads. Here is the exact process:
- Open your resume in Word or Google Docs and export it as a standard PDF.
- Open that PDF in any viewer.
- Select All (Ctrl/Cmd + A), copy, and paste into a plain text editor like Notepad or TextEdit.
- Read the pasted text top to bottom.
If it reads in logical order, you are fine. If it looks like the example below, your columns or tables are scrambling the text layer.
Scrambled (two-column resume):
Skills Experience Python Senior Software Engineer, Acme
Java 2021 to Present SQL Led migration of 12 services
Redis reducing deploy time from 25 min to 6 min
Correct (single-column resume):
Senior Software Engineer, Acme | Jan 2021 - Present
- Led migration of 12 services to Kubernetes, cutting deploy time from 25 min to 6 min.
Skills: Python, Java, SQL, Redis
The scrambled version interleaves your skills header with your job title and dates. No ranking model recovers cleanly from that.
A minimal ATS-safe template
Jane Doe
San Francisco, CA | jane@email.com | 555-123-4567
linkedin.com/in/janedoe | github.com/janedoe
EXPERIENCE
Senior Software Engineer, Acme Corp | Jan 2021 - Present
- Cut p99 API latency 40% (820ms to 490ms) via Redis caching and query batching.
- Led migration of 12 services to Kubernetes, reducing deploy time from 25 min to 6 min.
SKILLS
Languages: Python, Go, Java, SQL
Infrastructure: AWS, Kubernetes, Terraform, Redis, PostgreSQL
Plain, single column, real text, standard headings. It is not pretty, and that is the point.
Keywords and tailoring, the engineer's version
Keyword matching is real but often misunderstood. The ranking step looks for evidence that your history matches the job description, so the skills and phrasing in the posting should appear in your resume where they are true. If the role asks for "distributed systems" and "Kubernetes" and you have done that work, use those exact terms rather than a clever synonym.
Two practical moves:
- Mirror the job's language for real skills. If the posting says "gRPC" and you wrote "RPC services," add the specific term. Do not keyword-stuff skills you cannot defend in an interview.
- Match the seniority signal. A staff-level posting rewards scope words (led, owned, defined) and cross-team impact; a new-grad posting rewards shipped projects and internships. Jobscan's guidance is to tailor to each posting rather than sending one generic resume everywhere.
One caution we will stake a position on: do not put white/hidden keyword text on the page. AI screening tools increasingly flag or ignore it, and a recruiter who spots it in the parsed text will reject you outright.
What changed with AI screening in 2026
AI screening is common at large, high-volume employers but it is not universal, and the mechanics vary by vendor. The useful development for candidates is transparency. Greenhouse, one of the most widely used ATS platforms, publishes candidate-facing guidance on how it uses AI, and offers features like resume anonymization intended to reduce bias in early screening. That tells you two things: the parsed text is what the AI works from (so parsing quality still comes first), and demographic signals are increasingly stripped, so your content has to carry the argument.
Our take: the AI layer raises the value of specificity and lowers the value of decoration. A model summarizing your resume for a recruiter can only summarize what parsed cleanly. Clean structure plus quantified bullets is exactly what both a parser and a language model reward.
Once the resume does its job and lands the interview, the bar shifts entirely to how you solve and explain problems. That is where you should spend the rest of your prep time: running realistic mock interviews and drilling the patterns a target company favors. If you are aiming at a specific employer, our company breakdowns show what actually gets asked, from array and string warmups to mid-level graph work like "Find Duplicate Subtrees" and staff-level problems such as "Integer to English Words" and "Number of Ways to Stay in the Same Place After Some Steps."
FAQ
Do I need a different resume for each job?
You need one strong base resume and light tailoring per role. Adjust the summary line, reorder skills to match the posting, and make sure the exact technologies named in the job appear where they are true. You do not need to rewrite your whole history for every application.
Is a PDF or a Word document safer for ATS?
A text-based PDF exported from Word or Google Docs is safe with modern parsers and preserves your layout. Avoid PDFs that are actually images or screenshots, since they have no text layer to parse. If a specific application explicitly requests .docx, give it .docx.
Will a two-column resume automatically get rejected?
Not automatically, but it is the most common cause of scrambled parsing. Jobscan notes that columns and tables make many parsers read straight across the page and merge unrelated content. Run the copy-paste test; if your two-column layout reads cleanly as plain text, it is fine, and if it does not, switch to single column.
Does keyword matching mean I should list every skill?
No. List skills you can defend under questioning, and mirror the job description's exact terms for the ones you genuinely have. Padding your skills section with buzzwords you cannot back up gets exposed in the phone screen, which is a worse outcome than a slightly shorter list.
Sources
- Why ATS Tables and Columns Break Your Resume Parsing (Jobscan)
- Anatomy of an ATS-Friendly Resume Format (Checklist) (Jobscan)
- How to Convert Your Resume to ATS Format (Jobscan)
- The State of Resume Parsing: Does ATS Read Two-Column Resumes? (Enhancv)
- How Does Greenhouse Use AI? Everything Candidates Need to Know (Greenhouse)
- Anonymize Resumes (Greenhouse Support)