How to tailor your resume to a job posting
Tailoring a resume does not mean writing a new one. It means changing the emphasis, wording, ordering, and selection of your existing experience so the qualifications one particular role cares about are easy to see — without adding anything your history doesn’t support. University career centers give the same core advice: customize the resume for each posting rather than sending one generic version [1][2].
The method, in five steps
- Separate must-haves from nice-to-haves. Read the posting and list the requirements it repeats, puts first, or phrases as hard qualifications — those drive the screen. A typical posting orders its sections by importance: qualifications, then responsibilities, then overview [1].
- Match each must-have to real evidence. For every requirement on your list, find the strongest accomplishment in your history that demonstrates it, and lead with that. If two bullets could open a role description, the one that answers this posting’s first requirement goes first.
- Mirror the posting’s keywords — only where true. Recruiters search their applicant tracking system for the posting’s own terms, so describing the right experience in different words can keep it from surfacing [3]. Use the posting’s phrasing verbatim where your experience genuinely supports it, and stop there.
- Leave unsupported requirements out. Never add a claim you couldn’t defend in an interview. A requirement you don’t meet is better handled as an honest gap — addressed in the cover letter, an adjacent-experience framing, or a plan to close it — than as an invented line that fails its first follow-up question.
- Review every line before you send. Read the final document against your real history, line by line. This is also the step that catches over-eager rewording: if a bullet now implies more than you did, scale it back.
What actually changes
Five mechanisms, all within your existing experience:
- The summary is rewritten to lead with the experience this role asks for.
- Bullet wording shifts to the posting’s terminology where your experience genuinely matches it.
- Emphasis and ordering change: the most relevant accomplishments move up, less relevant ones shrink or drop.
- Skills are regrouped so the posting’s must-haves are visible at a glance.
- Keywords from the posting appear where true.
What does not change: the facts. No new employers, dates, titles, numbers, or skills.
Where the ATS fits in
An applicant tracking system (ATS) is the software most employers use to store, search, and filter applications; recruiters query it for the posting’s terms and knockout requirements rather than reading every submission [3]. Two practical consequences:
- Keyword alignment matters, truthfully. Use the posting’s own terms where they honestly describe your experience. Stuffing unmatched keywords is counterproductive: it fails the first interview question about them, and screening increasingly reads context rather than counting words.
- Simple formatting parses best. Single-column layouts without tables or graphics are read in order by ATS parsers; keep a plain-text version for application forms that mangle uploaded files [3].
- No tool or format guarantees a specific ATS outcome or an interview — screening configurations differ per employer and change over time. Treat any such promise skeptically, including from tools.
If you use AI to tailor: the fabrication problem
Generative language models are documented to produce fluent content that is not grounded in their input — the research literature calls this hallucination [4]. Applied to resumes, that failure mode is specific and costly: a rewrite can acquire skills, numbers, or accomplishments you never had, and a fabricated claim can end an interview or surface in a background check.
The mitigation is provenance: require that every rewritten line be traceable to a line in your own source resume, and that unsupported requirements be listed rather than filled in. Whether you do the check by hand (step 5 above) or use a tool that shows a trace per line, the standard is the same — if a claim has no source in your history, it doesn’t ship.
A worked example
From a real posting (Research Engineer, Interpretability at Anthropic, via its public Greenhouse listing) and a sample resume for a fictional platform engineer:
The posting asks for: “Profiling and optimizing ML training jobs, including multi-GPU parallelism and memory optimization.”
Evidence in the resume: “Profiled a distributed training job and fixed a host-memory bottleneck, cutting step time 22%.” (Stripe, 2022 – Present)
In the tailored resume: the bullet moves up, the headline becomes “Distributed Systems & GPU Infrastructure”, and the line carries its trace: source: Stripe, 2022 – Present.
Left out, with the reason: “Hands-on PyTorch/CUDA on GPUs or JAX/XLA on TPUs” — the background covers GPU cluster operations and training-job profiling but documents no PyTorch, CUDA, JAX, XLA, or TPU development, so the requirement is listed as unmet instead of written in.
Sources
- University of Arizona, Center for Career Readiness — Tailoring Your Resume.
- West Virginia University, Career Services Center — How to Tailor Your Resume to Impress Employers.
- Tulane University, School of Professional Advancement — Understanding Applicant Tracking Systems.
- Ji et al., Survey of Hallucination in Natural Language Generation, ACM Computing Surveys, 2023.