CVParsa for Search CommitteesJump-start your faculty search.
CVParsa understands and verifies the whole applicant pool in minutes, and returns an overview spreadsheet listing every applicant with selected verified information.
Rows shown as placeholders.
Every applicant, treated the same way.
- Same fieldsOne spreadsheet format, whatever the CV's layout, length, discipline, or country.
Every column in the spreadsheet
- Name
- Last and first name, in separate columns, and email
- Current position
- The applicant's current title, institution, and academic unit
- Doctorate
- The highest degree held, with its institution, field, year, and advisor
- Publications
- Total, first-author, and last-author counts, plus journal, proceedings, chapter, and book counts
- Interests
- Research keywords and a short characterization of the research program
- Source file
- The PDF each row came from, so every row traces back to its CV
- Same vocabularyDegrees, institutions, and publication types are mapped onto controlled academic vocabularies, so applicants are comparable across institutions and countries.
What gets standardized
- Degrees
- Hundreds of acronyms, spelled-out degrees, and institution-specific variations are resolved to one canonical vocabulary
- Institutions
- Each institution is kept as the CV names it and also resolved to a canonical name, so a pool can be sorted and grouped cleanly
- Publications
- Every entry is placed in a single publication-type taxonomy. The counts cover journal articles, proceedings, chapters, and books, with preprints linked to their version of record rather than double-counted
- Same checksEvery publication list is checked against Crossref and other authoritative bibliographic and library databases.
What the checks do
- On a match
- The published record replaces the CV's version: corrected titles, venues, and dates, and complete author lists. First- and last-author counts come from those author lists
- Without a match
- The entry keeps what the CV states and is plainly labeled unconfirmed. Absence of confirmation is reported as exactly that, not as absence of the work
- Precision first
- The checks are tuned to avoid false matches: an entry is never matched to a guess
So applicants are compared on their records, not on how their CVs happen to be formatted.
What's in your spreadsheet.
One row per applicant, and every job includes every field. You can leave out any field you don't want before you download.
| Field group | What it contains | What CVParsa decides |
|---|---|---|
| Person | Name, email, current position, institution, and department | Which position is the current one, and the institution's standard name |
| Education | Doctorate, doctoral institution, field, year, and advisor | Which degree is the doctorate, written in one standard form |
| Publications | Total, first-author, and last-author counts; journal, proceedings, chapter, and book counts | What counts as a publication, the applicant's place in the author list, and each entry's type |
| Interests | Research keywords and a short summary | The applicant's main research themes, drawn from the whole CV |
| Applicant pool | A one-page infographic of the whole pool | How the pool breaks down, at a glance |
Counts you can rely on.
The same CV can yield very different counts depending on who does the counting. CVParsa reads every entry, decides what it is, and checks it. On one faculty CV, 114 listed entries become 94 publications, with preprints and a patent kept separate.
Checking is about getting the numbers right, not judging the applicant. An entry that can't be confirmed isn't false; it may simply be missing from the databases. It stays in the record, marked unconfirmed.
Layout illustration, not real data.
The whole pool,at a glance.
Every job also includes a one-page infographic of the applicant pool: how many applied, their current positions, their doctoral fields and institutions, their publication and authorship profiles, and their research themes.
How it works.
Upload
ZIP one PDF per applicant, CVs or full dossiers, in any major language.
CVParsa does the reading
It understands and verifies every CV the same way, in about 10 minutes per 100 faculty CVs.
Download
Choose the fields to keep, then download the spreadsheet and the pool infographic.
See it work on a real CV
One CV's journeyFrom its first page to the finished, verified record. Plays automatically.
Click the slides or use the arrow keys to step through at your own pace. Open full screen · Full walkthrough
Why not just ask ChatGPT or Claude?Reading a CV is the part general-purpose chatbots already do well.
| Task | General-purpose chatbot (ChatGPT, Claude) | CVParsa |
|---|---|---|
| Read one CV and summarize it | Does this well | Does this well |
| Apply the same fields and vocabulary to every CV in a pool of hundreds | Output varies between runs | One schema, fixed academic vocabularies |
| Know that a fellowship title can be a postdoc, and that a CAREER award is both an honor and a grant | Inconsistently | Applied the same way to every CV |
| Count publications | Depends on how the question is asked | Publications counted by type; preprints and patents kept separate |
| Confirm each record against Crossref, NSF, NIH, and Cevian Labs databases | Not systematically, and prone to inventing citations it cannot find | Every record states which authority confirmed it, or that none did |
| Handle confidential applicant files | Consumer accounts raise data-governance questions | Encrypted, never used to train AI models, originals deleted within 30 days |
Built for confidentialapplicant files.
Simple, flat pricing.
One price per job, set by how many CVs you upload. Every job runs the full process, with no tiers to choose and nothing to unlock.
Every job includes the spreadsheet, the pool infographic, publication checking, and field omission.
Multi-use and site licenses are available, including tax-exempt invoicing for institutions. licenses@cevianlabs.io
Questions.
What is the output?
A spreadsheet with one row per PDF, describing each applicant, plus a one-page infographic of the whole pool. Every job includes the full field set:
- Applicant details: name, email, current position, institution, department, doctorate, doctoral institution, advisor, field, year, and research interests.
- Publications: total, first-author, and last-author counts, and journal, proceedings, chapter, and book counts, checked against Crossref and other authoritative bibliographic and library databases.
- Applicant pool: a downloadable infographic summarizing the full pool.
Can I hide some fields in the spreadsheet?
Yes. You can leave any field out of the output, for instance doctoral institution or email address. You choose these before downloading the spreadsheet.
Can I submit one big PDF containing many CVs?
No. CVParsa can't split a combined PDF. Each applicant needs their own PDF file.
Can I submit Word or other non-PDF files?
No. CVParsa works with PDF files only at this time.
Which languages and academic fields does CVParsa support?
CVParsa understands CVs in most major languages and recognizes international institutions and venues. Results are always returned in English.
All academic fields are supported. CVs in fields without standard publication formats (for example, artistic performances) may return partial publication data.
Are multi-use or site licenses available?
Yes, including tax-exempt invoice payment for institutional partners. Contact licenses@cevianlabs.io.
How does CVParsa handle confidential information?
Files are encrypted with AES-256 at rest and TLS in transit, and are never used to train AI models. Original files are kept at most 30 days, for customer support only.
How accurate is CVParsa?
Academic CVs vary widely in format across fields, and CVParsa is built to handle that variation. No automated tool is perfect.
In testing across hundreds of CVs in all disciplines, CVParsa achieves approximately 99% accuracy. In a batch of 100 CVs, you may see a few dozen extraction errors across all extracted fields combined.