Do I need to know LaTeX to use this?

No. You fill in an ordinary form — name, roles, bullets, education, skills — and the server writes the LaTeX for you. You never type a backslash. The reason it is worth using LaTeX at all is on the output side: the same input always produces the same page, text never reflows because a font is missing on someone else's machine, and the spacing rules that keep a parser reading your CV in the right order are enforced by the document class rather than left to you. If you do want to keep editing by hand afterwards, download the .tex and open it in the LaTeX editor on this site or in Overleaf.

Is it free, and how many CVs can I build?

Free, with no card and no trial. Without an account you get 1 CV per day. Signing in — also free — raises that to 5 per day, and verifying your WhatsApp number raises it to 15. Each of those is a compile, not a download: once a CV is built you can re-download the PDF, the .tex and the .zip as often as you like. A compile that fails never costs you a build. The day bucket resets at 00:00 UTC, which is 5.30 a.m. in Sri Lanka.

Do I really get the LaTeX source?

Yes, and it is the reason this tool exists. Download the .tex on its own, or the .zip, which also contains induwaracv.cls and the TrueType fonts the document uses. That bundle compiles unmodified on Overleaf or on a local TeX install — the class looks for a fonts/ folder beside the document, which is exactly how the zip is laid out. Almost no other online CV builder hands over the source, because the source is the thing that stops you needing them next time.

Is my CV stored on your server?

No. The form autosaves into your own browser's local storage and nowhere else, which is why there is a visible 'clear saved data' button rather than a buried setting — on a shared or library machine that button is the whole privacy story. Server-side, the compile runs in a throwaway sandbox directory that is deleted when the run finishes, and the response is marked no-store so nothing between us caches it. The only thing written to the database is one anonymous counter row recording which template and region were used. No name, no email, no CV text.

Will this guarantee I get past the ATS?

No, and be suspicious of anything that says it will. What the tool does is narrower and more useful: after your PDF is built it extracts the text back out of that PDF and shows you the literal characters a parser recovers, so you can see for yourself whether your email address, your job titles and your section headings survive. There is no score out of 100, because a score would be invented. The widely quoted '75% of CVs are rejected by robots' figure is not a measured rejection rate — it traces back to employer survey data in the Harvard Business School / Accenture 'Hidden Workers' report (2021), which asked employers what they think their own filters do.

Should I put a photo, NIC number and date of birth on my CV?

It depends entirely on where you are applying, which is why the region setting changes the form. For Sri Lankan employers a photograph, NIC, date of birth, civil status and two non-related referees are still conventional, and some application processes ask for them outright. For the US, UK, EU, Canada and Australia, leave every one of them off. It is not that attaching a photo is illegal — it is that those employers are barred from deciding on the characteristics a photo reveals, so many recruiters have a policy of discarding CVs that include one rather than carrying the risk. Build the local version, switch the region, and rebuild for the overseas application.

Match it to how much you have actually done: one page as a student or new graduate, two once you have a few years of relevant work, three only for academic or research CVs where the publication list is the point. The 'never more than one page' rule is folklore. In the CareerBuilder / Harris Poll survey of 1,138 US hiring managers (2018), typos were named a deal breaker by 77%, while a CV longer than two pages was named by 17% — the length rule is real but it is nowhere near the biggest thing you can get wrong.

Why does the tool offer durations instead of date ranges?

Because there is unusually good evidence for it. Kristal, Nicks, Gloor and Hauser ran a preregistered field experiment with 9,022 real job applications and published it in Nature Human Behaviour in December 2022. CVs that showed how long each role lasted ('3 years') instead of exact date ranges received roughly 15% more callbacks for applicants who had an employment gap, and about 8% more for applicants who did not. Switching the date mode rewrites every entry at once. If an employer explicitly asks for month-and-year dates, switch back — a rule in a job ad beats a field experiment.

Which template should I pick?

Clarity if you are unsure — single column, ordinary headings, no icons or tables, and the safest structure for automated parsing. Colombo for Sri Lankan employers, which adds the optional personal details block and a referees section on the same parse-safe skeleton. Compact Technical if your projects say more than your job titles, which is the usual case for students and developers. Scholar for academic applications, where publications, teaching and grants matter and length is expected. You can switch templates at any point without retyping anything — the data is separate from the layout.

Documents · Free · No signup

LaTeX CV Maker — an ATS-friendly CV, and the .tex source to keep

Fill in a form and this builds you a real LaTeX-typeset CV: a PDF compiled by an actual TeX engine, the .tex source and fonts to keep, and a read-back of the exact text an applicant tracking system pulls out of your file. Free, no account needed for your first CV.

Source included, not locked in Shows the extracted text, not a fake score Nothing you type is stored

By Induwara Ashinsana— Executive Director, Ryzera TechnologiesUpdated Jul 30, 2026

Pick a starting point

All four are single column with plain headings, because that is what survives being read by a machine. You can change template at any time without retyping anything.

ATS-safe structure

CVs built with this tool

How it works

From a form to LaTeX

You never type a backslash. The form collects structured data — a contact block, one entry per role, bullets, education, skills — and the server turns that into a LaTeX document. Every string you typed is escaped on the way in, because LaTeX reads % & # $ _ \as instructions: a bullet reading “grew revenue 40% & cut cost” would otherwise truncate at the percent sign with no warning. That escaper is also why the tool refuses LaTeX you wrote yourself. A document that can call \input can read files off the server and print them into your PDF, so the generator is the only thing here allowed to write TeX commands.

From LaTeX to a PDF

The source compiles with Tectonic (XeTeX) inside a sandbox with no network access and a working directory that is deleted the moment the run ends. A warm compile takes about a second, which is what makes previewing as you type practical. If the compile fails you get the engine log and a plain-English reading of it — and a failed build never costs you one from your daily allowance.

The document class does the parse-safety work

Advice pages tell you to avoid tables and icons. Here those rules are compiled in, so no template can opt out of them by accident: ligatures off, hyphenation off, no header or footer on page one, dates set below the role instead of flush right, and minipages instead of tables wherever two things sit side by side. Each rule came from extracting text out of a compiled PDF and seeing what came back wrong, not from repeating what everyone else writes. The section below lists what each one prevents.

Then the PDF is read back

Once your PDF exists it is pushed through a text extractor and you are shown the literal characters that come out: whether your name is genuinely the first thing read, whether your email extracts as youremail rather than merely “an email address was found”, whether every section heading survived and in what order. There is no score out of 100, because a score would be invented — and the widely repeated claim that three quarters of CVs are auto-rejected by robots is not a measurement either. It comes from asking employers what they think their own filters do.

You keep the source, and nothing is stored

Three downloads: the PDF, the .tex on its own, and a zip holding the .tex, the induwaracv.cls class file and the TrueType fonts — a bundle that compiles unmodified on Overleaf or a local TeX install. You can also send the source straight to the online LaTeX editor and keep working there. Meanwhile the form autosaves into your own browser and nowhere else; the server keeps no copy of your CV, and the only database write is one anonymous row recording which template and region were used, so the counter above has something honest to count.

What actually breaks a CV when a machine reads it

Almost never the wording. It is the layout, and the failures are silent — the PDF looks perfect on screen while the text coming out of it is wrong. These are the 7 that matter, with the mechanism behind each and what this tool does about it.

Words come out joined and unsearchable

Most TeX fonts render "ff", "fi" and "ffi" as single ligature glyphs (U+FB00 and friends). A search for "Officer" then returns zero matches against a CV that visibly reads Chief Financial Officer.

Ligatures are disabled for the whole document (Ligatures=NoCommon), so every letter extracts as itself.

Keywords split in half at a line break

Justified text inserts real hyphens, and those hyphens survive extraction: "internationaliza-tion" never matches "internationalization".

Hyphenation is switched off entirely and the text is set ragged-right instead of justified.

Something other than your name is read first

Parsers overwhelmingly treat the topmost text block as the candidate name. Anything in a page header — a document title, a page number, your own name repeated — extracts before it and hijacks that heuristic.

Page one carries no header or footer at all, so the first extracted line is your name.

Dates get detached from the job they belong to

Right-aligning a date with a wide whitespace gap trips the column-detection heuristic in PDF text extractors, which can orphan every date to the very end of the extracted text. Switching to a table does not fix it — the gap itself is the cause.

Dates sit directly below the role on their own line, never flush right.

Two columns interleave into gibberish

A tabular layout extracts row by row, so a sidebar and the main column alternate mid-sentence. This is the single most common reason a good-looking designer CV reads as nonsense to a machine.

Side-by-side regions use minipages, and the recommended templates are single-column throughout.

A phone number with nothing to identify it

Contact icons are font glyphs. They extract as an unrelated private-use codepoint or as nothing, leaving a bare string of digits the parser has to guess about.

Icons are off by default; contact details carry plain text labels.

No readable text at all

A CV exported as an image, screenshotted, or flattened by a converter has no text layer. This is the one universally documented hard failure — SmartRecruiters returns an explicit UNPARSABLE_RESUME error and Oracle Recruiting Cloud does no OCR of its own.

A TeX-generated PDF always has a real text layer, and the panel shows you the extracted text as proof.

The advice you have read, checked against its source

Every one of these is repeated as fact by pages currently ranking for CV advice. 3 of them do not survive being looked up. 2 do, and those are the ones worth acting on.

Myth

“75% of CVs are rejected by ATS robots before a human ever sees them.”

There is no published measurement behind that percentage. It traces back to the Harvard Business School / Accenture report 'Hidden Workers: Untapped Talent', which surveyed employers about their own hiring systems and found most executives believed their filters were excluding qualified candidates. That is employer belief about their configuration — not a measured rejection rate for your file. What is true, and enough to act on, is that large employers do parse CVs automatically and that a file the parser cannot read is a documented failure mode.

Source: Fuller, Raman et al., Hidden Workers: Untapped Talent — HBS / Accenture, September 2021

Overstated

“Recruiters spend exactly 6 seconds looking at your CV.”

The number comes from a 2012 eye-tracking study run by the job board TheLadders, which observed 30 recruiters during an initial screen; the company's 2018 follow-up reported 7.4 seconds. Small, vendor-run and never peer reviewed. It is a decent reason to make the top of page one immediately legible. It is not a measurement anyone should quote to two significant figures.

Source: TheLadders, Eye Tracking Study, 2012 (n = 30)

Overstated

“Use Google's official resume formula: Accomplished [X] as measured by [Y] by doing [Z].”

Good advice, wrong attribution. Laszlo Bock posted it on LinkedIn on 29 September 2014 as his personal formula. He was Google's SVP of People Operations at the time, which is how it became 'the Google formula', but Google has never published it as a policy or a standard. Use the shape because quantified achievements are independently supported by hiring-manager survey data, not because a company mandated it.

Source: Laszlo Bock, LinkedIn, 29 September 2014

Holds up

“Typos will get your CV thrown out.”

This one holds up. In a CareerBuilder / Harris Poll survey of 1,138 US full-time hiring managers and HR professionals, 77% named typos or bad grammar as a deal breaker — the single most common answer. An unprofessional email address was named by 35%, a lack of quantifiable results by 34%, and a CV longer than two pages by only 17%. Proofread before you obsess over length.

Source: CareerBuilder / Harris Poll, 2018 (n = 1,138)

Holds up

“An employment gap sinks your application.”

Gaps do cost callbacks, and there is a preregistered field experiment on how to reduce that: 9,022 real applications, published in Nature Human Behaviour in December 2022. Showing how long each role lasted ('3 years') rather than exact date ranges raised callbacks by roughly 15% for applicants with a gap and about 8% for applicants without one. That finding is why this tool has a duration date mode at all — no mainstream CV builder offers one.

Source: Kristal, Nicks, Gloor & Hauser, Nature Human Behaviour, December 2022 (n = 9,022)

The four templates

Same parse-safe skeleton underneath all four. What changes is which sections lead, which fields exist, and the typeface.

Clarity

Everyone — the recommended starting point

Single column, boring section headings, no icons, no tables. The safest possible structure for automated parsing — and the default for that reason.

Colombo

Sri Lankan local employers and public-sector applications

Sri Lankan conventions: optional NIC, date of birth and civil status, plus a referees section. Same parse-safe structure underneath.

Compact Technical

Developers, students and new graduates

Dense one-pager that puts projects and skills above experience — built for people whose work speaks louder than their job titles.

Scholar

Academics, researchers and postgraduate applicants

Academic CV: unlimited length, publications as plain text entries, teaching and grants. No bibliography toolchain required.

A Colombo CV and a London CV are not the same document

This is the part global CV builders get wrong for Sri Lankan users. A photograph, an NIC number, a date of birth, civil status and two named referees are ordinary on a local CV — some application processes ask for them outright. Send the same file to an employer in the US, UK, EU, Canada or Australia and every one of those lines is a liability, because those employers are barred from deciding on what that information reveals and many recruiters simply set the file aside rather than carry the risk. Switching the region rewrites the form, so the two versions come from one set of facts instead of two drifting files.

Field	Sri Lanka	US / UK / EU / CA / AU	Why
Photograph	Conventional, often expected	Leave it off	US, UK, EU, Canadian and Australian employers may not decide on the characteristics a photo reveals, so many recruiters discard photo CVs rather than carry the risk.
NIC / national ID number	Commonly asked for	Never include it	Overseas it serves no screening purpose and is a plain identity-theft risk on a document you email to strangers.
Date of birth or age	Conventional	Leave it off	Age discrimination is unlawful in each of those jurisdictions, so volunteering the information helps nobody and can get the file set aside.
Civil status, gender, nationality	Still common on local formats	Omit — state work authorisation instead if it is relevant	"Eligible to work in the UK" answers the only question an employer actually needs answered.
Referees with contact details	Expected — ads often ask for two non-related referees	Omit, or say references are available on request	Overseas, references are requested at offer stage. Either way, ask your referees before you print their phone number.
Length	Two to three pages is normal, especially with referees	One page early career, two with several years' experience	Academic CVs are the exception everywhere and run as long as the publication list requires.

Three worked examples

Three people, three regions, three sets of settings — and in each case the bullet that changed and the lines the extraction panel reported back.

Example 1 · Colombo template · Sri Lanka

A UCSC graduate applying in Colombo

Sanduni, BSc (Hons) Information Systems, final year, applying for an Associate Software Engineer role advertised by a Colombo 03 software house that asks for two non-related referees.

Settings that mattered

Education moved above Experience — two internships do not outrank a degree yet
Personal block on: date of birth, NIC and civil status, as the application form asks for them
Referees section with a lecturer and her internship supervisor, with permission asked first
Date mode left on ranges, because nothing in her history has a gap to smooth over

Before

Did my final year project on an e-channelling system for clinics.

After

Built an e-channelling web app (Next.js, PostgreSQL) piloted by 3 private clinics for 6 weeks; cut average booking time from about 4 minutes on the phone to 40 seconds online across 210 real bookings.

What the extraction panel said

Text layer present — 431 words recovered across 2 pages
Your name is read first — "Sanduni Rajapaksha"
Email extracts correctly
Sections read in the right order — EDUCATION → EXPERIENCE → PROJECTS → SKILLS
63% of your bullets contain a number

The personal block and the referees are conventional here and leaving them out can read as an incomplete application. Keep this build as the local version — then switch the region to International and rebuild before sending anything abroad, which strips all five personal fields in one click.

Example 2 · Compact Technical template · International

A Sri Lankan developer applying overseas

Ruwan, four years of backend work in Colombo, applying to a remote-first company in Berlin and a startup in Austin. He took eight months off in 2024 to care for a parent.

Settings that mattered

No photo, no NIC, no date of birth, no civil status — the region setting removes the fields entirely
Date mode set to durations, so roles read "3 years" and "1 year" rather than exposing the 2024 gap as a hole between two ranges
Projects placed above Experience — his open-source work is stronger than his job titles
Icons left off, so the phone number and email carry their labels into the extracted text
Referees dropped; nobody outside South Asia expects contact details on the CV itself

Before

Improved checkout performance and fixed database issues.

After

Cut p95 checkout latency from 1.9s to 380ms by replacing 14 N+1 ORM queries with a single materialised view; the change held at 12× normal traffic through the December sale with no rollback.

What the extraction panel said

Text layer present — 388 words recovered across 1 page
Your name is read first — "Ruwan Fernando"
Phone number extracts
Sections read in the right order — PROJECTS → EXPERIENCE → SKILLS → EDUCATION
71% of your bullets contain a number

Two applications, one master file. Because you keep the .tex, the Berlin version and the Austin version are the same source with the region and the summary changed — not two documents drifting apart in a downloads folder.

Example 3 · Scholar template · United Kingdom

An academic CV for a postdoctoral application

Nuwan, a lecturer applying for a two-year postdoctoral position in the UK, where the publication list is the substance of the application and length is expected.

Settings that mattered

Publications entered as finished citation strings — no BibTeX file, no bibliography toolchain, no compile errors at 2 a.m.
Custom sections added for Teaching, Grants and Invited Talks
Education above everything, which is the academic convention
Length left alone: four pages is normal here

Before

Published several papers on hydrology and taught undergraduate modules.

After

Sample citation entry, typed as plain text: Perera, N., & Silva, K. (2025). Rainfall-runoff modelling for the Kelani basin under CMIP6 scenarios. Sri Lanka Journal of Water Resources, 12(2), 44-59.

What the extraction panel said

Text layer present — 1,642 words recovered across 4 pages
Your name is read first — "Nuwan Perera"
Sections read in the right order — EDUCATION → PUBLICATIONS → EXPERIENCE → SKILLS
4 pages — longer than typical for this much experience

That last line is a warning the tool shows and you should ignore. The length budget is calibrated to industry CVs; an academic CV with a real publication list is supposed to run long. Every check in the panel is advisory, and the panel tells you which mechanism each one is about so you can decide when it does not apply to you.

Frequently asked questions

Related tools

LiveDeveloper

LaTeX Editor

Write LaTeX in the browser and see the compiled PDF as you type. Runs a real TeX engine in an isolated sandbox, so there is nothing to install and your document is never stored. Includes a quick reference, worked examples and plain-English explanations of the usual errors.

Open tool

LiveFreelance

AI Cover Letter Generator

Generate a tailored, one-page professional cover letter for Sri Lankan job seekers from a job title, pasted job ad or skills, and experience level — with editable paragraphs and a copy-ready letter.

Open tool

LiveAI

AI Resume ATS Checker

Paste a resume and a job description to see an ATS score, missing keywords, format issues parsers struggle with, and section gaps. Scoring follows Workday, Greenhouse, and Lever guidance plus Jobscan's 2024 threshold. Runs in your browser, no signup.

Open tool

Sources

Fuller, Raman et al., Hidden Workers: Untapped Talent — Harvard Business School / Accenture, September 2021. Employer survey data, and the actual origin of the “75% auto-rejected” claim.
Kristal, Nicks, Gloor & Hauser, Reducing discrimination against job seekers with and without employment gaps — Nature Human Behaviour, December 2022. Preregistered field experiment, n = 9,022.
CareerBuilder / Harris Poll, 2018 — survey of 1,138 US full-time hiring managers and HR professionals on CV deal breakers.
Laszlo Bock, LinkedIn, 29 September 2014 — the “Accomplished [X] as measured by [Y] by doing [Z]” formula, posted as personal advice rather than a Google policy.
TheLadders eye-tracking study, 2012 (n = 30 recruiters) — the origin of the “6 seconds” figure.

Rate this tool

Be the first to rate

Comments & feedback

Spotted a bug or want an improvement? Tell us — our team reviews every comment, and good ideas get built. Comments are public and anonymous.

Pick a starting point

How it works

From a form to LaTeX

From LaTeX to a PDF

The document class does the parse-safety work

Then the PDF is read back

You keep the source, and nothing is stored

What actually breaks a CV when a machine reads it

The advice you have read, checked against its source

The four templates

A Colombo CV and a London CV are not the same document

Three worked examples

Frequently asked questions

Do I need to know LaTeX to use this?

Is it free, and how many CVs can I build?

Do I really get the LaTeX source?

Is my CV stored on your server?

Will this guarantee I get past the ATS?

Should I put a photo, NIC number and date of birth on my CV?

One page or two?

Why does the tool offer durations instead of date ranges?

What actually breaks a CV when a machine reads it?

Which template should I pick?

Related tools

LaTeX Editor

AI Cover Letter Generator

AI Resume ATS Checker

Sources

Comments & feedback