thinkbridge Academy admin
Add users, manage cohorts, author the curriculum, audit who did what. Close this tab when you're done — your sign-in token lives only in this browser tab.
Sign in
Use your PocketBase superuser email and password. (Admin users created via the cohort flow sign in via the magic link they received — once you've clicked it, this page will load you straight into the admin shell.)
Cohorts needing attention
Active cohorts, busiest first. Jump straight into a cohort's cut, screen, or roster.
Add a user
Users
Find anyone here. To manage who's in a cohort, use the Cohorts tab.
Create a cohort
Cohorts
Each cohort runs one track. Open a cohort to see and manage its roster.
Submissions
Browse every member's work by Day — what the thinkschooler submitted, the AI verdict, and how the mentor scored it. Pick a cohort, then a Day/piece.
Pick a cohort and a Day/piece.
Biweekly cut
Pick a cohort — the cut auto-selects based on how many interview rounds have happened (cut 1 after round 1, cut 2 after round 2, …) and the drop threshold defaults accordingly (50 / 60 / 70 / 75 / 80), which you can override. Composite ranking — passed · interview · quality · pace · mentor, each scored 0–100 (absolute; a component with no signal yet is excluded and the rest reweighted). Interview is a recency-weighted average across every round held so far — the latest round counts most, so the score tracks trajectory, not a single interview. Each round's scores are also calibrated for interviewer severity (every rater re-centered to that round's mean, so a lenient or harsh interviewer doesn't move a candidate); the raw average is shown in grey alongside. Anyone strictly below the drop threshold is flagged as a recommendation; nothing is auto-dropped. Review, adjust, add a reason, and confirm. Dropped members lose access immediately and every drop is audit-logged.
Pick a cohort and load the ranking.
Interview coverage & assignment
Who's interviewing whom, and who's scored this cycle. Assign interviewers to candidates (panel-capable — many per candidate), remove an assignment, and spot the un-interviewed. Independent of the ranking above — pick a cohort + cycle here and load.
Load the cohort's interview coverage.
Item banks
The Aptitude Screen draws a balanced random 42 from the screen bank per candidate; the optional Build With AI prep course (at /learn/) draws its mastery checkpoints from a separate course pool, disjoint from the screen bank. The Business Software Engineer track is a full cohort curriculum (its pieces live on the track). All three seeds are idempotent — re-run after editing the source in the repo.
Re-seeding the BSE, thinkOne Sales, or Direct the Build track is authoritative: it rebuilds the curriculum and removes any pieces it no longer defines. Safe when no one has started the track; check for in-progress learners first.
Other cohort curriculum tracks — additive (adds any new pieces by slug, never removes existing ones):
Screen results & interview assignment
Who has taken the screen, their machine score, whether they cleared the bar, and any integrity flags (advisory — a person reviews them with the live interview). Filter the funnel and assign passers (including flagged-but-passing) to interviewers — one at a time or in bulk; assigned interviewers show as named chips you can remove with the ✕. This is the same assignment the interviewer view reads. Candidates never see their score.
Graduation & certification
Each member's evidence summary — composite, pieces passed by category, competencies. A mentor recommends an outcome; you finalize it, which freezes the snapshot and mints a certificate. The bar is the same composite threshold; in a certify cohort, clearing it = ready to certify.
Pick a cohort and load.
Add a track
A track is a curriculum. Cohorts pick a track; pieces are scoped to it. The plan below is what this track's members see at /plan/.
Tracks
Add a category
Curriculum
The selected track's curriculum — categories in order, with their pieces. You're here from Tracks (or a cohort's View curriculum); switch tracks with the selector, or use ‹ Back to Tracks.
Competency coverage
Loading…
Add a competency
All competencies (grouped by bucket)
Audit log
Keys & tokens
Integration keys the platform uses, stored server-side in the secrets collection. Values are write-only — you can set or replace a key here, but it is never shown back. Every change is recorded in the audit log (key name + length only, never the value).