Sample report

This is a synthetic resume — not real data. Your own report will be customized to your actual resume and target.

Upload yours →

Your judgment

synthetic-resume.pdf

target: Senior Software Engineer at Stripe

Parser agreement
0%
3 parsers
AI-legibility
0/ 100
placeholder weights
Inter-modal δ
0.00
LLM ↔ ATS
Structural parse
ATS Report
Per-field agreement
  • name100%
  • email100%
  • phone100%
  • linkedin url95%
  • github url88%
  • personal urls67%
  • skills79%
  • education100%

Experience alignment: 83%

AI interpretation
AI Perception Report
4 of 4 LLMs responded
QueryMeanσ scalarρ reasoning
Seniority (1-10)6.500.50
0.12
Technical depth (1-10)7.000.70
0.18
Top strengths
0.31
Role fit (1-10)7.200.40
0.14
Final-round probability0.420.06
0.16
Key credential
0.22
Missing signal
0.28
AI-authored probability0.180.05
0.09
Cross-system agreement
Inter-modal δ
0.11
alignedmild gaplarge gap

How far the LLM read of your seniority drifts from what the ATS parsers extracted as your level. Soft signal — calibration improves in M5.

Top strengths · consensus
  • Quantified ownership of system-design work at scale (Meta TAO migrations)
  • Cross-stack range — proficient in both backend Go and frontend TypeScript
  • Demonstrable shipped product impact (B2B SaaS revenue retention)

Across surviving LLMs, ranked by mention frequency.

Missing signal · consensus

No public-facing technical writing, conference talks, or open-source contributions linked from the resume — for a Senior IC role at Stripe, evidence of external technical voice is a routine signal that's currently missing.

Most-detailed answer from any responding LLM.

Your judgment, in plain English

The narrative version

Plain-language explanation of the data above, with the specific numbers from your resume. The tables remain authoritative — this section translates them.

What the parsers saw

All three parsers extracted name, email, and phone identically and agreed perfectly on the education section. Disagreement clustered around personal-URL extraction (33% across parsers) and skills (21% Jaccard distance), suggesting the resume's contact block is robust but the skills section may read inconsistently to keyword-matching ATSes. ATS legibility (mean 78%) is solid; fragility (variance 0.04) is low — meaning all three parsers extracted similar fill rates, so structurally this resume is on firm ground.

Your experience bullets

Across 11 total bullets in the two experience roles, 7 (64%) contain quantified outcomes — a healthy quantification rate that beats the rough median for engineering resumes. 9 of 11 (82%) start with strong action verbs (built, shipped, led), and only 1 contained a vague phrase ('cross-functional'). The earlier Meta L4 role had stronger quantification (5/6 bullets) than the current SaaS role (2/5) — adding numbers to the recent bullets would tighten the narrative. Mean bullet length is 142 characters, in the readable range.

How the AIs read you

All four LLMs agreed closely on seniority (σ 0.5 — mean 6.5/10, mid-to-senior), with similar consensus on fit for Senior Software Engineer at Stripe (σ 0.4 — mean 7.2/10). Disagreement was higher on technical depth (σ 0.7) and the top-strengths list (ρ 0.31 — the four models pointed to slightly different dimensions of strength). Inter-modal δ of 0.11 means the AI judges' read of seniority is essentially aligned with the ATS-derived structural level — no wide gap between how the parsers and the LLMs see this candidate. Final-round probability mean of 42% reflects competitive but not stretched targeting.

Three things to fix
  1. Add quantified outcomes to bullets 2 and 3 of the current SaaS role — even rough numbers ('reduced page load time ~40%') anchor the bullet against AI-and-ATS readers alike. The Meta L4 bullets are a good template.
  2. Replace 'cross-functional' (the one buzzword detected) with a concrete description of which teams collaborated and what shipped as a result — this is the kind of specificity that improves both AI readability and human signal.
  3. The personal-URL section had highest parser disagreement (33%); audit your portfolio/scholar/blog links for consistent formatting (https:// prefix, no trailing slashes) so the parsers extract them identically.
Caveat

Placeholder weights — calibration against real outcome data lands in M5. The σ and ρ metrics on the right are direct measurements and do not depend on those weights — they are reliable today.