Evaluation · a real run · live providers · AS-1.1.0

Watch the Standard do its work.

This page is not a mock-up. Below is an actual dossier run — Chidinma Eze — fictional persona; real pipeline output — graded by the same tribunal that grades every dossier. Draft by draft, findings and all. Honesty is the aesthetic.

7/8artifacts passed on their final draft.
4/8passed on the first draft — we publish this so the repair loop can’t flatter itself.
14reports shipped. Every draft’s verdict is kept, including failures.

The Tribunal · independent review

Independent reviewers examining a candidate’s evidence dossier.

The repair loop, draft by draft

FAILresume_designeddraft 1craft mean 75.6

Hard checks — 9 passed

All deterministic checks passed.

Craft axes — Claude critic

Voice52
Specificity82
Quantification85
Positioning78
Tailoring68
Evidence honesty80

Repair brief — issued to the Forge

[craft:voice] Phrases like 'directly addressing latency tuning requirements' and 'demonstrating high-throughput payments experience' read as meta-commentary on the job brief rather than natural resume prose, making the voice feel mechanical and keyword-driven rather than confident and human. [craft:specificity] Strong concrete nouns (PostgreSQL connection pooling, p95, TypeScript, 5 engineers) ground each bullet in real technical detail. [craft:quantification] 38% latency reduction, 12,000 rps, and 5 mentees are all specific, plausible metrics with no signs of invention. [craft:positioning] Bullet order mirrors brief priority (latency/pooling, throughput, mentoring+migration), which is appropriate, but the justificatory clauses undercut the impression that this ordering reflects genuine career emphasis rather than brief-matching. [craft:tailoring] Covers three of four brief requirements explicitly, but Rust (the bonus skill) is entirely absent, and the migration bullet substitutes a TypeScript migration for anything payments/infra-related, which may or may not satisfy the 'team migration' criterion as intended. [craft:evidence_honesty] No unsupported superlatives; each claim stays within the bounds of what the stated metric can support. Rewrite bullets to state achievements plainly (e.g., 'Cut API p95 latency 38% via PostgreSQL connection pooling and query tuning') without appended clauses that explicitly reference the brief's language — let the facts demonstrate relevance instead of narrating it. If Rust experience exists, add a bullet naming the specific system/tool; if it doesn't, leave it out rather than implying coverage. Clarify whether the TypeScript migration bullet is the intended 'team migration' evidence for this brief, and if a more directly relevant migration (e.g., payments infra) exists, lead with that instead.
FAILresume_designeddraft 2craft mean 75.7

Hard checks — 9 passed

All deterministic checks passed.

Craft axes — Claude critic

Voice58
Specificity72
Quantification85
Positioning75
Tailoring66
Evidence honesty88

Repair brief — issued to the Forge

[craft:voice] Bullets follow a templated 'Verb + metric + tool' pattern with no distinct phrasing or personality — reads as generic resume prose rather than a specific engineer's voice. [craft:specificity] Numbers are present (38%, 12000 rps, 5 engineers) but the mechanics behind them are thin — no mention of which pooling tool (e.g., PgBouncer), what latency tuning beyond pooling was done, or what the TypeScript migration entailed (scope, stack, timeline). [craft:quantification] All three claims are backed by concrete figures; no vague qualifiers like 'significantly' or 'greatly' are used, which is good craft. [craft:positioning] Lead bullet correctly matches the brief's top requirement (Postgres pooling/latency), and payments scaling follows — solid ordering. Mentorship/migration bullet placed last, in line with lower brief priority. [craft:tailoring] Covers 3 of 4 required brief items (pooling/latency, payments scale, mentoring/migration) but the migration example (TypeScript) doesn't reinforce brief relevance beyond mentorship; Rust bonus is silently omitted rather than being addressed as absent, missing an opportunity to note it as a gap or non-issue. [craft:evidence_honesty] No inflated language or unverifiable superlatives; claims stay within plausible, measurable bounds. Rewrite each bullet with sharper technical specificity and a more confident, individual voice: name the pooling tool/technique used (e.g., PgBouncer, pool sizing, query tuning) and briefly note the latency-tuning method beyond pooling. For the migration bullet, clarify what was migrated and why it mattered to the team/system, so it reads as a deliberate leadership story rather than a generic line item. If Rust experience is genuinely absent, do not attempt to imply it — leave it out, but consider explicitly noting adjacent systems-level work (e.g., Postgres internals, concurrency tuning) to partially bridge that bonus qualification without fabricating claims.
PASSresume_designeddraft 3craft mean 85.2

Hard checks — 9 passed

All deterministic checks passed.

Craft axes — Claude critic

Voice68
Specificity88
Quantification90
Positioning82
Tailoring85
Evidence honesty90
FAILstory_bankdraft 1craft mean 69.8

Hard checks — 8 passed

All deterministic checks passed.

Craft axes — Claude critic

Voice48
Specificity60
Quantification85
Positioning65
Tailoring62
Evidence honesty85

Repair brief — issued to the Forge

[craft:voice] All three entries follow an identical 'Facing X, I did Y, achieving Z' template, reading as formulaic AI-generated prose rather than a distinct human voice. [craft:specificity] Mentions PostgreSQL pooling and TypeScript migration but omits concrete mechanics — e.g. which pooling tool (PgBouncer?), what tuning parameters were changed, or what the migration process/timeline looked like. [craft:positioning] Order roughly matches brief priority (pooling, scaling, mentorship) but the Rust bonus is entirely unaddressed, leaving a gap where the strongest possible differentiator could have been surfaced if evidence exists. [craft:tailoring] Covers 3 of 4 brief requirements solidly; the Rust systems programming bonus is not mentioned at all, so the artifact reads as generic backend/payments experience rather than one deliberately mapped to this brief's full scope. [craft:evidence_honesty] Claims are appropriately scoped to specific, plausible metrics (38% p95 reduction, 12k rps, 5 engineers) with no inflated language — no overreach detected. Rewrite the three stories with varied sentence structure instead of a repeated template. Add concrete technical detail to each: name the pooling tool/config used and what 'tuning' specifically meant; describe what made the TypeScript migration hard (scope, timeline, risks) and what mentoring concretely involved (pairing, code review, ramp-up plan). If any Rust systems programming experience exists, add a fourth story naming the project and outcome — if none exists, do not fabricate one; instead ensure the existing stories are positioned to make clear Rust is not part of this candidate's background rather than leaving a silent gap.
FAILstory_bankdraft 2craft mean 76.2

Hard checks — 8 passed

All deterministic checks passed.

Craft axes — Claude critic

Voice58
Specificity70
Quantification80
Positioning82
Tailoring75
Evidence honesty85

Repair brief — issued to the Forge

[craft:voice] All three stories follow an identical 'Facing X, I did Y, achieving Z' template, giving the bank a mechanical, resume-bullet cadence rather than a confident individual voice. [craft:specificity] Numbers are present (38%, 12000 rps, 5 engineers) but the technical detail behind the pooling fix (e.g., pooler tool, config change, connection limits) and the scale trajectory (from what baseline to 12000 rps) is missing, leaving the actions somewhat generic. [craft:quantification] Each story is backed by a specific figure (38% p95 reduction, 12000 req/s, 5 mentees) with no unsupported or vague claims — solid quantification throughout. [craft:positioning] Story order mirrors the brief's priority (pooling/latency, then scale, then mentoring/migration), putting the most relevant evidence first. [craft:tailoring] Directly covers 3 of 4 brief requirements (pooling/latency, payments scale, mentoring+migration) but never addresses the Rust bonus, and the migration story pairs TypeScript rather than a system relevant to the brief's stack, slightly diluting fit. [craft:evidence_honesty] Claims are framed at a believable, bounded strength (percentage improvement, explicit rps figure, explicit mentee count) with no inflation beyond what's stated. Rewrite the three stories with varied sentence structure and a more personal, confident tone (avoid repeating 'Facing X, I did Y' three times). Add one concrete technical detail per story (e.g., which pooling tool/library was used, what the pre-scaling baseline was, what the TypeScript migration entailed — scope, timeline, number of services) to sharpen specificity. If any Rust systems programming experience exists, add a fourth story addressing it explicitly since it's called out as a bonus in the brief; otherwise leave it out rather than implying it.
FAILstory_bankdraft 3craft mean 76.3

Hard checks — 8 passed

All deterministic checks passed.

Craft axes — Claude critic

Voice48
Specificity66
Quantification85
Positioning78
Tailoring80
Evidence honesty88

Repair brief — issued to the Forge

[craft:voice] All three stories follow an identical templated formula ('Facing X... Chidinma did Y' / 'When X threatened... Chidinma did Y' / 'To achieve X... Chidinma did Y'), making the prose read as generated rather than a distinct human voice recalling real events. [craft:specificity] Numbers are present but the mechanics are thin — 'introduced PostgreSQL connection pooling' doesn't name the pooler (e.g., PgBouncer), pool size, or what else was tuned; 'scaled the payments service' doesn't say what changed (sharding, caching, async workers) to reach 12000 rps; 'led the migration to TypeScript' doesn't state scope (services, LOC, timeline). [craft:positioning] Story order maps well to brief priority (pooling/latency first, throughput scaling second, migration+mentoring third), but Rust systems programming is not addressed at all — acceptable since not fabricated, though a brief note on stack context would help tailoring. [craft:tailoring] Covers three of four brief requirements directly (pooling/latency, scale to thousands rps, mentoring+migration) with no fabricated Rust claim, which is correct restraint but leaves that requirement fully unaddressed rather than acknowledged as absent. Rewrite the three entries with varied sentence structure so they don't share the same context-then-action template, and add one concrete technical detail per story (e.g., the pooling tool used, what change enabled 12000 rps, or the scope/duration of the TypeScript migration) to raise specificity without inventing new metrics.
FAILresume_atsdraft 1craft mean 76.2

Hard checks — 10 passed

All deterministic checks passed.

Craft axes — Claude critic

Voice52
Specificity68
Quantification86
Positioning80
Tailoring78
Evidence honesty84

Repair brief — issued to the Forge

[craft:voice] Phrases like 'directly addressing latency tuning requirements' and 'demonstrating high-throughput payments experience' read as meta-commentary aimed at the reviewer rather than natural resume prose, undercutting authentic voice. [craft:specificity] Pooling and migration bullets state outcomes but omit concrete mechanics (e.g., pool size/tooling like pgbouncer, what the TypeScript migration entailed, team size, timeline) that would make the actions feel earned rather than summarized. [craft:quantification] Strong use of hard numbers (38% p95 reduction, 12,000 rps, 5 engineers) that appear plausible and well-scoped to claims made. [craft:positioning] Bullet order follows the brief's priority (latency/pooling first, then payments scale, then mentoring/migration), which is appropriate, though Rust is not addressed at all, leaving one bonus criterion silent rather than positioned as absent. [craft:tailoring] Covers three of four brief requirements explicitly (pooling/latency, payments scale, mentoring+migration) but never signals the Rust gap, leaving the reader to infer no experience exists there. [craft:evidence_honesty] No detected overreach; each claim stays within a scope the stated action could plausibly support, and the artifact correctly avoids inventing Rust experience rather than stretching to cover it. Rewrite bullets in plain achievement-first language without meta phrases like 'directly addressing' or 'demonstrating experience' — just state the action and result. Add one concrete technical or scope detail per bullet (e.g., pooling tool/config used, what the TypeScript migration involved and its scale) to strengthen specificity without inventing new facts. Optionally add a short explicit note that Rust experience is not yet present, or omit silently but ensure this omission is a conscious tailoring choice rather than an oversight.
PASSresume_atsdraft 2craft mean 79.5

Hard checks — 10 passed

All deterministic checks passed.

Craft axes — Claude critic

Voice68
Specificity78
Quantification85
Positioning82
Tailoring70
Evidence honesty88

The parse-back diff — every field survived machine reading

The ATS résumé PDF from this run, re-parsed by Assay’s deterministic engine and diffed against the source profile. Fidelity: 100% · 0 fields lost.

FieldSource profileMachine read-backVerdict
nameChidinma EzeChidinma Ezesurvived
emailchidinma.eze@example.comchidinma.eze@example.comsurvived
exp0.orgPaystackPaystacksurvived
exp0.titleSenior Backend EngineerSenior Backend Engineersurvived
exp0.startYm2021-032021-03survived
exp0.endYmPresentPresentsurvived
exp1.orgAndelaAndelasurvived
exp1.titleBackend EngineerBackend Engineersurvived
exp1.startYm2018-062018-06survived
exp1.endYm2021-022021-02survived

verified against Assay's deterministic parser and ATS format law — not a simulation of any specific vendor.