JEV Resume Analyzer

Review a CV with fast, structured AI judgments, optionally against a job posting. Upload a PDF or DOCX, check the extracted text, and get a traceable report: every assessment shows the exact question, criteria, answer, and probability distribution behind it.

Built on TypeSafe (System One model, jev-latest): structured judgments composed in code, not prompt-and-parse text generation.

What it does

• Full-document CV review: work clarity, personal contribution, interpretability, outputs/results, scope, and optional-summary specificity, each as a typed judgment over the whole document.
• Optional job alignment: paste a posting; two document-level questions assess how the CV evidences relevant responsibilities and qualifications.
• Honest uncertainty: four answer outcomes (present, absent, not_applicable, unable_to_assess) with real probability distributions. "No summary" is not a defect; "unable to assess" is not a zero.
• Traceable findings: expand any finding to see the evaluated question, its criteria, the answer, and per-option probabilities. Explanations are predefined rubric text, clearly labelled: never generated advice.
• Privacy-first demo: no accounts, no database, no localStorage for CV text, no client-side keys. One Jev request per analysis; failed requests never produce fabricated reports.

Stack

Layer	Tech
Frontend	React 18, Vite, TypeScript, Zod-validated API contract
Backend	FastAPI, Pydantic, pypdf, defusedxml (DOCX XML traversal)
AI	TypeSafe System One API (jev-latest), one batched request per analysis

Quick start

Requirements: Python 3.11+, Node 20.19+ or 22.12+, uv, a TypeSafe API key.

# Backend
cd backend
cp .env.example .env
# Edit .env and set TYPESAFE_API_KEY
uv sync
uv run uvicorn app.main:app --host 127.0.0.1 --port 8017

# Frontend (second terminal)
cd frontend
npm ci
npm run dev
# open http://127.0.0.1:5273

The Vite dev server proxies /api to the backend on 127.0.0.1:8017.

Tests

# Backend: 69 synthetic tests + lint (live test is gated behind JEV_RUN_LIVE=1 and a running API)
cd backend && uv run pytest -q && uv run ruff check app tests

# Frontend: 30 tests, typecheck, build
cd frontend && npm test && npm run typecheck && npm run build

# Browser flows (needs Chromium; see docs/frontend-contract.md)
cd frontend && CHROMIUM_PATH=<chromium-binary> npx playwright test

Design boundaries

• Full-document assessments only; no fabricated bullet-level findings.
• No overall rating is shown yet: scoring maps stay null until calibrated on reviewed examples.
• No hiring predictions, ATS scores, truthfulness verdicts, or generated rewrites.
• Localhost demo: CV text is sent to TypeSafe for analysis; retention terms should be verified before any public deployment.

Docs

• docs/architecture.md: proposed architecture, API contracts, runtime flow
• docs/frontend-contract.md: binding frontend↔backend schema (implemented in frontend/src/api/schema.ts)
• docs/backend-implementation.md: backend verification evidence and contract decisions
• docs/frontend-contract.md and the rubric modules under backend/app/rubrics/ define what every question means

Status

Working end-to-end on localhost: upload → editable extraction preview → one batched Jev request → traceable report. Deferred: generated suggestions (phase 2), OCR, accounts, public deployment.