For residency, fellowship, and admissions review committees
Review every applicant. See the evidence. Make better decisions.
Evidence-linked. Transparent. Human-controlled.
RankLyst structures every letter, SLOE, and personal statement into the same evidence categories for every applicant — surfacing what a tired reviewer might miss on file forty, and tracing every flag straight back to the sentence that triggered it. Reviewers rank independently before seeing the committee's opinion, guarding against anchoring. RankLyst never makes the selection decision — it organizes evidence and reviewer judgment into a transparent committee rank list.
The problem
Hundreds of applications. A handful of reviewers. Thousands of pages.
Letters, personal statements, and files get read differently depending on who's reading them, how many they've already read that day, and how tired they are by file forty.
RankLyst turns that unstructured pile into a standardized, evidence-linked review — the same categories, the same rigor, applied to every applicant, every time.
See it in action
What your committee actually sees
Real screens from the live product, from an in-progress pilot dataset — not mockups.
From our pilot dataset
One flag, traced back to the sentence that raised it
A real screen from the live product, on a synthetic applicant from our own test dataset — not a real candidate.
How review changes
From eight documents to one page of evidence
An illustration of the idea, not a live screenshot — every row below links back to an actual sentence in an actual document inside the real product, the same way the flag example above does.
Applicant #147
ERAS application, MSPE, four letters, personal statement, research history, SLOEs, transcript, test scores.
A reviewer opens all eight documents and starts hunting for what matters.
Applicant #147 — overall assessment
And here's the real thing
Not a mockup — this is the actual product
A real screen from the live product, on a synthetic applicant from our own test dataset — not a real candidate. Same idea as the illustration above, built from data the product already extracts.
Why it's worth switching
Review faster, more consistently, with the evidence in view
Review faster
Surface what matters across hundreds of applications without opening eight documents per candidate to find it.
Review consistently
The same categories, applied the same way, whether it's the first file of the cycle or the four-hundredth.
See the evidence
Every score, flag, and assessment traces back to the actual sentence in the actual letter or application.
How it works
Four steps, the same ones your committee already runs
Upload what you have
Reference letters and application files, per candidate — combined files are fine, RankLyst sorts out the rest.
Structured extraction
Competency ratings, exam scores, and narrative flags are pulled out and cross-checked against the letter's own rating scale.
Independent committee review
Each reviewer ranks and comments before seeing anyone else's — anchoring only happens after everyone's already weighed in.
Your committee decides
Disagreements, rank lists, and side-by-side comparisons inform the conversation. The final call is always human.
Two seats at the table
Fairer review looks different depending where you're sitting
RankLyst is licensed by the organization and used by the committee — but every mechanism below exists because of what it means for the candidate on the other side of the file.
Defensible decisions, less second-guessing
- A documented, consistent process you can stand behind if a decision is ever questioned.
- Committee disagreement surfaces early, in the review, instead of in the post-decision debrief.
- Nothing new to teach candidates — you're reviewing the same letters and files as always.
Your file, read on its own terms
- Every reviewer ranks you independently before seeing a single other opinion — no one's first impression sets the tone for the room.
- A strong, well-earned letter can't get quietly averaged down into "mediocre" because a reviewer misread the scale.
- Race, ethnicity, disability status, and citizenship are never extracted or scored — not hidden after the fact, never touched at all.
What RankLyst does
Structural bias resistance, not a UI toggle
Every mechanism below is enforced in how data is stored and shown — not a setting a reviewer can accidentally skip.
Independent ranking, first
Your committee ranking is locked in before you can see anyone else's — so the first opinion in the room isn't the one everyone anchors to. Our own computed reference tier follows the same rule: hidden until you've ranked, and it only resurfaces if your read lands far from it.
Blind mode, on demand
Hide school and writer institution with one toggle when your committee wants to review on file content alone.
Every flag traces to a source
Deterministic checks and AI-read phrases are labeled separately, so a reviewer always knows whether a flag is a hard rule or a judgment call.
One check for what no checklist can ask
A general, always-on read for what falls outside every specific requirement — an unexplained gap, a timeline that doesn't add up, a credential rare enough to matter. Every note is a verbatim quote from the file itself, never folded into any score. We held the bar deliberately high: a flag on every file trains a reviewer to stop reading it, so this only speaks up on the rare file that genuinely earns it. Freeform AI tools surface this by accident, when they surface it at all; here it's a standing, quote-backed check, held to the same bar on every file.
Disagreement, not a blurred average
Norm-referenced composite scoring flags outliers and reviewer disagreement instead of quietly averaging a strong letter into a mediocre one.
Protected characteristics: never
Race, ethnicity, disability, citizenship, and other protected characteristics are never extracted into a structured field, flag, or score — excluded at the schema level, not filtered after the fact.
Encrypted, with a real retention policy
Application materials are encrypted at rest. Set a retention window and flag stale files for deletion — reviewed by an admin, never automatic.
Rank list & side-by-side comparison
A numeric rank order compiled from your reviewers' independent input, plus a facts-only comparison view for the 2–4 candidates your committee is weighing head-to-head. Once your shortlist is set, generate one interview packet for just those candidates — every flag and strength already on file, nothing your coordinator has to re-type by hand.
Works alongside your process
Upload the reference letters and application files you already have. No new candidate portal, no change to how people apply.
Why we built this
"We're GME committee members ourselves. We know what it's like to be hours and a dozen files into a review session — tired, aware our attention wasn't what it was for the first applicant, and increasingly sure our own read was being swayed by whoever on the committee had spoken first. That didn't feel like a fair evaluation for every candidate, and we didn't think it had to be this way. We built RankLyst because we believed there was a better way to run this process — one that catches what a tired read misses and protects every reviewer's independent judgment before the room starts talking."
Ken Jacobsohn, MD, FACS, DipABLM
Founder
Trust & security
Security your records office can actually verify
No vague assurances — here's exactly what protects your applicants' data today.
Encrypted at rest
Every applicant record and letter is stored in an encrypted database, not just protected in transit — the data on disk is encrypted, full stop.
Role-based access, enforced
Admins manage uploads, accounts, and deletions. Reviewers can view files and comment — they can't touch the account list or delete a record. It's built into the schema, not a setting.
A full audit trail
Every upload, role change, and deletion is logged. When a record is deleted, that history survives — who, when, why — even though the applicant's own data does not.
Retention with a human in the loop
Set a retention window and RankLyst flags exactly which records are past it. Nothing is ever deleted automatically — an admin reviews the list and confirms before anything is permanently removed.
Your own deployment, not a shared database
Each institution runs its own instance with its own database. Your applicant data never sits in the same table as another institution's.
Account security built for real threats
Passwords are salted and hashed (PBKDF2, 390,000 iterations) — never stored in plain text. An account locks automatically after repeated failed logins.
Why committees choose it
Easy to roll out, easy to feel the difference
Add your files, and go
Upload what your reviewers already have — no new candidate portal, no migration project, no IT ticket. Most committees are reviewing their first file within minutes of getting access.
Get your time back
Ratings, scores, and flags are extracted and organized automatically. Reviewers spend their time on judgment calls — not retyping numbers off a PDF or cross-referencing a scale by hand.
Bias reduction that isn't optional
Independent ranking, identity redaction, and protected-characteristic exclusion are enforced in how data moves through the system — not a setting a reviewer can forget to turn on.
Where it fits
Built for any structured, multi-reviewer file review
The mechanism is generic — independent ranking, identity redaction, and norm-referenced scoring don't care what the file is. Here's where it's proven today, and where it's headed next.
RankLyst Talent adapts the same evidence-linked approach to resumes and CVs — every requirement match tied to a verbatim quote, ranked consistently, evidence attached instead of an AI opinion you have to take on faith. Pilot-stage today; reach out at hello@ranklyst.app to see it against a search you've already run.
What RankLyst doesn't do
The boundaries are as deliberate as the features
Decision-support tools earn trust by being honest about their limits.
Early feedback
What early reviewers are telling us
RankLyst is in active pilots now. Real quotes from real committees are coming as programs finish their first cycle — nothing fabricated, nothing placeholder-dressed-as-real.
Quote from a pilot reviewer — coming soon.
RESERVED FOR A REAL QUOTEQuote from a program administrator — coming soon.
RESERVED FOR A REAL QUOTEQuote from a committee chair — coming soon.
RESERVED FOR A REAL QUOTEPiloting RankLyst and willing to be quoted? Tell us — we'd rather have three real sentences than thirty invented ones.
FAQ
Questions committees actually ask
Is the AI making the selection decision?+
Does this touch protected-characteristic data?+
Where does our data go? Is it appropriate for regulated data?+
What specialties or fields does it support?+
Can one reviewer try it, or does my whole committee need to sign on?+
How is it priced?+
I'm a candidate, not a reviewer — does this affect me?+
Is this only for residency and fellowship selection?+
Who's behind RankLyst?+
Get in touch
Request access
Tell us a little about your committee or your own review — we'll follow up directly to set up an account and walk through your data questions.
hello@ranklyst.app