Features

A tour of what Resume Studio does today. Most of these are visible the moment you open the app — nothing here is buried behind config files.

Editing

• Dual-language side-by-side editing. Every translatable field renders as two inputs at once. Pick any two of your supported locales as primary and secondary, swap them with one click, or hide the secondary column when you want to focus.
• Translation assist on every field. Each secondary input has a Copy from primary button (no network) and, when a translation provider is configured, a Draft translation button that pre-fills a machine translation for you to review. Drafts are always flagged as review-required.
• Multiple translation providers. LibreTranslate (Docker-managed local instance with a pick-your-languages install, or a remote URL you host), DeepL, Google Cloud Translation, Microsoft Azure Translator — or the AI model you configured for assist, with zero extra setup. Switch between them from Settings.
• Re-detect languages. A refresh button in the language switcher scans your content and adds any locale it finds to your supported list — handy after importing a CV.
• Profiles with their own competencies. Write several profiles — each a tag line plus a short and a long summary — and let every Resume View present one. The profile’s tag line becomes that view’s resume title. Each profile owns an ordered set of key competencies that travel with it: a view shows exactly the competencies of the profile it presents, in the order you set on the Profile page (drag to reorder, or pull in a batch of existing ones with checkboxes). Competencies live in a shared library, so the same one can belong to several profiles when you want to reuse it — and the Key Competencies page has a By profile view that groups them under each profile.
• Course, certification and presentation dates & categories. Group courses and certifications with a shared set of categories (an editor-only organizing aid, never exported) and filter by them while editing. Courses and presentations take a from/to date range like your other timeline sections — so a talk you’ve given regularly over a period reads correctly.
• Rich text where it matters. Project and role descriptions support bold, italic, underline, and bullet/numbered lists.
• Profile photo and company logo. Uploaded, downscaled in-browser, embedded in exports.
• Drag-and-drop reordering on every section, with keyboard up/down buttons retained for accessibility.
• Undo / Redo with debounced history (Ctrl/Cmd+Z, Ctrl/Cmd+Shift+Z).
• Global search (Ctrl/Cmd+K) across every section, skill, role, and the header — jump straight to any item.
• Career timeline. An overview chart of employments, education, and projects, with work-history-gap detection and a full-width zoom.
• Cross-language drift check. The Overview flags fields whose two languages have drifted apart — a number that changed on one side but not the other (a wrong headcount, a dropped percentage, a timeline that runs a year longer in one language), or prose that grew in one language while the other stayed a stub. Click a flag to jump straight to the field. It’s the natural companion to the completeness meter: one tells you what’s missing, the other what’s out of sync.
• Freshness warnings. The overview flags expired or expiring certifications and suspiciously long “ongoing” items; anything you’ve checked can be snoozed for a year.

Skills, roles & categories

• Shared registries. Skills, roles, and industries live once and are referenced everywhere — rename a skill and every project updates.
• Registry merge. The “Løsningarkitekt” vs “Løsningsarkitekt” problem: pick one as canonical, the other gets rewritten everywhere it’s referenced and removed.
• Skill categories & showcase. Group skills into your own categories (drag-and-drop between them), highlight the ones to showcase, and the exported “Skills Showcase” section builds itself from those choices.
• Curated skill library. Autocomplete against 1,200+ canonical skill names so imports and typing don’t mint near-duplicates; related-skill suggestions; one-click auto-categorization of uncategorized skills — offline, no service involved.
• Skill matrix. An exportable skill × years × proficiency × last-used table derived from your project history.

AI assist — bring your own model

One model powers every AI feature, and you choose where it runs: a local Ollama (the app can run it in Docker for you — your CV never leaves the machine), OpenAI, Anthropic, Google Gemini, Mistral, or any OpenAI-compatible endpoint (LM Studio, Groq, OpenRouter, …). The model field offers a live pick-list of what your provider actually has. Every AI button states plainly whether content stays on this computer or goes to your provider, results are always drafts you review before they touch your CV, and with no model configured the buttons simply hide. Every feature also has a manual path — copy a generated prompt into whatever AI you already use and paste the answer back — so nothing requires an API key.

• One-line summaries. Draft a short description from a long one — per field, or “Bulk summarize” for a whole section at once (with a confirmation that explains what will run).
• Job-posting tailoring. Paste a posting and get a proposed view: section detail levels, item exclusions, a drafted intro, and a gap list.
• Skill suggestions. Propose the skills a project’s prose demonstrates, matched against your existing registry so it links “React” rather than minting “React.js”.
• Drafted project highlights from the project description.
• Strengthen the wording. Coach an existing description into tighter, stronger prose — grounded in what you actually wrote, never inventing employers, numbers, or claims. It sees the entry’s own fields (name, dates, issuer…), so it won’t restate them or ask you for facts that already have a field — and text that already reads well is told so instead of reworded.
• Assists keep working while you do. Start one and carry on — move to another item, another section, or reload the page. The spinner is still there when you come back, a finished suggestion waits until you accept or discard it (ticks and all), and a notice tells you the moment it’s ready wherever you happen to be.
• Anonymization check. Scan an anonymized view for real client names that leaked through in prose.
• Cover-letter draft. Turn a job posting plus the CV you’re sending into a drafted letter body — grounded in your real experience, never invented.
• Page-fit advice. When a view runs over its page limit, get concrete suggestions for what to cut.
• AI import and bulk add — see the Import section below.

The advanced assists

If the model you configured is a strong one, tick “This is a high-end model” in Settings and a second tier appears. These read the whole CV and make comparative judgements about it — the kind of work a small local model answers fluently and wrongly, which is why you declare it rather than the app guessing. With the box unticked they are simply not there.

• Review my whole CV. A prioritised list of what a reader would hold against it: descriptions with no outcome, wildly uneven detail, repeated phrasing, timeline gaps. Every finding links straight to the item.
• Consistency & voice. One register across the document — same person, same tense, technology names spelled the same way — as a batch of before/after rewrites you tick to apply.
• Cross-language meaning check. Does the Norwegian actually say what the English says? Catches dropped sentences and terms translated three ways.
• Achievement mining. Finds outcomes buried in long prose and proposes them as highlights or competencies — quoting the sentence that supports each one, and filling both language columns.
• Job fit report. Requirement by requirement against a posting: what you can evidence, what you can’t, and what you nearly can — the gap you can close honestly by naming something you already did.
• ATS keyword audit. Reads the text a view actually exports and reports which of a posting’s terms appear in it. The first pass needs no AI at all. Its most useful verdict is “in your CV but excluded by this view” — fixed by re-including an item, with no writing.
• Profile generator, view introductions, and “what’s missing” per section.
• Registry hygiene. Proposes skill merges and categories for the uncategorised tail. Proposal-only: nothing is pre-ticked, every row states what it deletes and how many references it rewrites, and a confirm names the totals first.
• Cover-letter angles and critique.

Runs survive you walking away from them: results are kept per CV, so you can go and fix the thing a finding just told you about and come back to the rest.

Teams and accounts (self-hosted)

The desktop app never asks you to sign in — it is your machine, and a login screen would be friction and nothing else. A self-hosted instance can hold several people instead.

• Setting it up is one code. The server prints a one-time setup code when it starts with no accounts. Spend it, and the account you create becomes the owner and takes ownership of every résumé already there.
• Invite by link. The owner generates a single-use link and passes it on however they already talk. No email server required.
• Your résumés are private by default. Nobody else on the instance sees them until you share one — and sharing makes it readable, never editable. A colleague opening a shared CV gets it read-only; they cannot change your words.
• The owner can see everything. Deliberate, and stated plainly here because it is the sort of thing you should not discover later: it is what makes staffing work, whole-instance backups, and recovering a CV after somebody leaves possible.
• Sign in with your username or your email — whichever you remember.
• Four ways back in if you forget your password: a recovery code you saved when the account was made, a link the owner mints for you, an email link if the owner has configured mail, or — for an owner locked out with nobody above them — a command run on the server itself.

Email is optional and off unless the owner configures it. It is used for exactly two things: a password-reset link and confirming an address. No CV content is ever emailed, and an address only receives resets after its confirmation link has been followed.

Multi-resume

• One app, many master CVs. Keep separate CVs for different lines of work, joint ventures, or career chapters. The picker is the home screen; each CV has its own URL, history, and supported languages.
• Per-resume language pair. Each CV remembers which two languages you were last editing in.
• “Who knows what” skill matrix. With more than one CV in the instance, the picker offers a skill × person grid across everyone — who has which skill and at what proficiency, with a filter for the skills more than one person shares (team overlap) versus the ones only a single person holds (bus-factor risks). Click a name to open that CV.
• Shared registries across CVs. From the same panel, “Share registries across resumes” links matching skills, roles and industries to one shared registry — after which renaming a skill in any CV updates it in all of them, while each person keeps their own proficiency and highlights.

Resume Views — targeted exports

• Curated subsets of the master CV. A view names a set of sections to include, items to exclude, a starred-only toggle, and a custom intro.
• Purpose note. Jot down why a view exists — which client, tender, or role it’s for — as a private reminder shown on the view (with an edit pencil). It’s never exported.
• Per-section detail levels. Each section can be Off, Summary, Tabulated (aligned columns), or Full — so a one-pager and a deep technical CV share the same source data. Sections can also flip to starred-only individually and be bulk-selected by type facets (e.g. only Research publications).
• Per-view styling. Density, body size, fonts, heading and accent colors, page margin, tag style, item dividers and bullets, section icons, custom section headings, date formats, per-section sort and summary layout — all stored on the view, not the master CV.
• Configurable header and footer. Choose which contact fields appear, the labels and separators, name/title type size, photo and logo placement, and a footer note.
• Live preview pane. The document re-renders as you tune the view, with a page-count estimate against your page limit.
• Export templates. Named presets (compact technical, formal management, minimal one-pager) that seed a view’s style, header/footer, and section detail in one click.
• Job-posting tailoring. Paste a job posting and run it with your configured model in one click — or copy the generated prompt into any LLM and paste the answer back. Either way the view reorders and trims itself for that role, as a proposal you review first.
• Page limit with real advice. Set a page budget, watch the live page-count estimate, and ask the AI what to cut when you’re over.
• Anonymized variants. A per-view toggle that anonymizes customers and redacts references to initials — for tenders and broker submissions.
• Per-view export language — the same view can ship in English to one client and Norwegian to another.
• Promoted Projects and Skill Matrix as synthetic sections — surface starred projects or a skills table without restructuring the master.

Cover letters

• A letter per application, paired with a view. A cover letter is its own document that references the Resume View it accompanies — write several against one CV, one per role you apply for.
• Drafted from the posting. Paste the job posting and draft the letter body with your configured model (or copy the prompt into any LLM), grounded in the CV you’re actually sending — it won’t invent employers or numbers.
• Matching letterhead. The letter borrows the linked view’s fonts and your contact details, so letter and CV read as one submission. Export to PDF, DOCX, or plain text, in any of your languages.

Export

• What you previewed is what you send. The live preview, the PDF, the Word file and the ATS text state the same facts and honour the same style choices. Optional extras — links, team size and allocation, referee contact details, grades, expiry dates, locations — are switched on per view, so what an export contains is something you chose rather than a side effect of which button you pressed. Every group starts off.
• PDF — a one-click vector download, rendered from the same section catalog as the preview.
• DOCX (.docx) via the docx library, lazy- loaded so the bundle only grows when you actually export.
• Plain text & Markdown — ATS-friendly exports for application portals that mangle formatted documents.
• Europass XML — the SkillsPassport format EU and Norwegian public tenders ask for, and the round-trip partner of the Europass import. Covers identity, work history, education and language skills (with CEFR levels); the sections Europass has no concept of stay in the richer PDF/DOCX exports.
• Fully localized output. Every piece of document chrome a client reads — section headings, month names, “Present”, contact-field labels, skill-matrix columns, language levels — ships translated in all 15 offered languages, so a Norwegian or German export never leaks English labels.
• Language proficiency done properly. Spoken languages render as a compact one-liner or a full Europass CEFR passport (A1–C2 per skill), your choice per view.

Import & backup

• CVpartner JSON import. The importer handles both shapes CVpartner emits (object and interleaved-array localized values), normalises int → en, scans content for under-declared locales, and links projects to work experiences through the source IDs.
• LinkedIn import. Drop the LinkedIn data-export .zip on the picker and get a working resume.
• Europass import. Reads both SkillsPassport XML and Europass profile JSON.
• AI-assisted import from PDF or Word. Paste your CV’s text and run the import with your configured model in one click — or download the instruction template, run it in any LLM, and paste the JSON back. Either way you preview the result before it becomes a resume.
• Per-section bulk add. Paste raw source material (an old CV chapter, a course list, a project log) and turn it into many items in one reviewed batch — with the same one-click-or-manual AI choice, and duplicates detected against what the section already has.
• Portable JSON backup. Per-resume export from the editor; versioned format with a migration scaffold so older backups keep loading.
• Server-side snapshot history. Every save is snapshotted (last 50 per resume, duplicates skipped). The header’s History button restores any version — and the restore itself is undoable.

Persistence & offline tolerance

• Auto-save to a local SQLite database (debounced ~1 s).
• localStorage fallback per resume, so a momentary outage never costs work. Edits flush the moment the server returns.
• Status visible in the header — Saving / Saved / Offline / Queued / Conflict, with a count of any resumes still waiting to sync.
• Optimistic concurrency. If two devices race on the same resume, the loser sees a non-blocking Conflict modal with a labelled diff and a keep-mine / discard-mine choice.

Cross-computer sync (desktop)

• Backup folder in your existing cloud sync (Google Drive, Dropbox, OneDrive). Resume Studio writes one JSON file per CV there, atomically, and merges newer content back in continuously while it runs — not only at launch — so edits made on another computer land within seconds of your sync client delivering them, even if you leave the app open for days. If the CV you’re viewing was updated elsewhere, a small “updated on another device — Reload” notice appears.
• One file per person, on purpose. A CV is one identified individual’s data, so removing someone from your backups is deleting one file — not editing a document that holds everybody. You can also hand a single CV to someone by sending its file.
• Newest-wins per CV. A restore drops a snapshot first, so it’s reversible from History. Nothing is removed except CVs you deleted yourself: a deletion writes an id-and-timestamp marker the other machines honour, and a copy saved after the deletion counts as a revival and is kept.
• No real-time multi-writer. Designed for one person hopping between computers, not for two people editing the same CV at once.
• Automatic updates. The desktop app checks GitHub Releases daily (or on demand from the tray / Settings) and installs a new version in place with one click — no reinstall, your data untouched.

Privacy & security posture

• Your CV never leaves your machine unless you point translation or AI assist at a remote provider — and the app says exactly where content goes before you run anything. Local options (Docker LibreTranslate, Ollama) keep everything on your computer.
• No account, no telemetry, no analytics.
• Loopback-only on desktop. The local server binds 127.0.0.1 — the app is never exposed to your network, including when you reach it under the friendly local name (resumestudio.localhost) instead of the IP.
• Auth-gated when self-hosted. Server deployments require a bearer token; the browser exchanges it for an HttpOnly session cookie so the token never lives in JavaScript-readable storage.
• Content sanitised at the render boundary. Rich text, view styling, and imported view configs are sanitised before they reach the export or preview pipelines.

The full detail — every host the app can contact, what each provider receives, how sync folders and deletion work — is in PRIVACY.md. To report a security issue, see SECURITY.md.

---

Ready to try it? Head to the downloads.