CV documents

A document is a .tex file you own. vitakit renders content/ into cv-tex/generated/<document>/, one file per section; latexmk typesets it. There is no -shell-escape and no Python: LaTeX only typesets what is already on disk.

pnpm

pnpm build:cv                 # every document
pnpm build:cv cv              # one document
pnpm build:cv --no-pdf        # only generate .tex (no TeX installation needed)

npm

npm run build:cv                    # every document
npm run build:cv -- cv              # one document
npm run build:cv -- --no-pdf        # only generate .tex (no TeX installation needed)

After build:cv --no-pdf, a plain latexmk cv.tex in cv-tex/ works as well, e.g. from an editor or a CI action: cv-tex/latexmkrc lets LaTeX find the generated document class.

Two files per document

cv-tex/<name>.tex	order and structure — moving a line moves a section
cv-tex/<name>.yml	selection — what goes in, optional

\documentclass{vitakit-cv}
\vitadocument{cv}                          % selects generated/cv/
\addbibresource{../content/publications.bib}
\input{preamble.tex}
\input{generated/cv/profile.tex}           % header data from profile.yml

\begin{document}
\vitaheader

\vitasection{employment}
\vitasection{education}
\vitasection{teaching}

\section{Publications}\vitainput{publications}

\vitasection{supervision}
\vitasection{talks}
\end{document}

# cv-tex/cv.yml
visibility: cv            # include public and cv entries, not private
sections:
  talks:
    where: { kind: invited }
    limit: 10

Macros

Macro	Meaning
\vitadocument{name}	which generated directory to read
\vitaheader	name, tagline and contact line from profile.yml
\vitasection{id}	heading + the generated entries for that section
\vitainput{id}	the generated file only, no heading or list (publications)
\vitasectiontitle{id}{Text}	heading wording for this document
\vitaentry{date}{title}{context}, \vitanote{…}	what generated files emit

Because \vitasection emits a real \section, \tableofcontents and PDF bookmarks work as usual.

Selection options

visibility: public | cv | private     # default: cv
sections:
  <id>:
    where: { kind: [invited, seminar] }  # field equals this value, or one of them
    since: 2020                          # start year at least this
    until: 2024                          # start year at most this
    limit: 10                            # keep the newest N
    title: Selected talks                # heading wording, this document only
publications:
  groups: [published, preprint]          # omit the software group

That is the whole list, and it is meant to stay that way. where compares against the field as written (for rich-text fields, their source text; for an institution, its name). There is no query language, no or, no negation: selection subtler than this belongs in a hand-written .tex or an ejected template, which is already how everything else specific is done here.

Unknown keys are an error, naming the file — so a typo is never silently ignored.

Making a variant

A document is a file, so make one by copying a file:

cp cv-tex/cv.tex cv-tex/teaching-uni-x.tex
cp cv-tex/cv.yml cv-tex/teaching-uni-x.yml      # optional

Change \vitadocument{cv} to \vitadocument{teaching-uni-x} in the copy, reorder the \vitasection lines, adjust the YAML, then build:cv teaching-uni-x.

The data is untouched, so your website and every other variant are unaffected.

A prescribed order

Calls often prescribe exactly what a dossier must contain. That is purely a matter of the .tex:

\begin{document}
\input{prose/application-xyz-cover.tex}   % 1. cover letter
\newpage
\tableofcontents                          % 2. table of contents
\newpage
\vitasection{publications}                % 3. publications
\input{prose/research-statement.tex}      % 4. research statement
\vitasection{teaching}                    % 5. courses taught
\vitasection{supervision}                 % 6. supervised theses
\input{prose/teaching-statement.tex}      % 7. teaching statement
\end{document}

Use \vitasectiontitle{teaching}{Courses taught} in the preamble when the call prescribes the wording too.

Prose

Cover letters and statements live in cv-tex/prose/*.tex as plain LaTeX, and are never read by the website build. Include them with \input wherever you want them.

Styling

preamble.tex is yours. Generated files only use the macros above, so redefining those changes the design without touching anything vitakit produces:

\renewcommand{\vitaentry}[3]{\item[#1]\textsc{#2}\par{\small #3}}

For deeper changes, vitakit eject vitakit-cv.cls copies the class into cv-tex/, where it is yours and takes precedence, and vitakit eject talks.eta copies the template that generates a section.

Do not edit cv-tex/generated/. Everything there — the section files, the profile header and the copy of the class — is rewritten on every build. An ejected file always wins: cv-tex/ comes before the generated directory in the LaTeX search path.

A template whose name matches no section is an error rather than a file that is silently never read — so talk.eta tells you it meant talks.eta.

What the build writes

cv-tex/
  cv.tex, cv.yml, preamble.tex, prose/   your files, never modified
  cv.pdf                                 the result
  generated/cv/                          everything else: section .tex files,
                                         vitakit-cv.cls, and all of latexmk's
                                         auxiliary output

cv-tex/generated/ is disposable: delete it whenever a build looks stale.

Bibliography

The class loads biblatex, and \vitainput{publications} prints one list per group. The groups are the website's: vitakit decides which entry belongs where (an entry without status is published, and Published counts as published) and generated/<document>/publications.tex selects each group by citation key, so the PDF and the website always list the same entries. arxiv becomes an eprint link.

For a list of your own, \printbibliography[keyword=published] also works: the status field is copied to keywords, as written. Pass \documentclass[nobib]{vitakit-cv} to load biblatex yourself instead.

Building the bibliography needs biber (part of every current TeX distribution); latexmk runs it for you.