Radar Loker & Lab (7-in-1 Suite)
Seven privacy-first, deterministic client-side web tools built with zero backend dependencies: Indonesian Job Scam & Red Flag Detector, 30-Day AI Job-Application Telemetry Dashboard, Thesis Defense Simulator, Fresh-Grad Salary vs UMK Benchmark, HRD CV Roaster, Datasheet Register to STM32 HAL / ESP-IDF C Driver Generator, and Labor Law / PKWT Contract Diff Explainer.
What I built
A unified 7-in-1 static web application suite deployed on Vercel (radar-loker.vercel.app) and GitHub Pages, divided into four career-protection tools and three technical/legal laboratory tools:
- 01 — Detektor Loker Red Flag (
/index.html): Deterministic 18-rule scanner scoring Indonesian job postings from0–100across illegal diploma retention (SE Menaker No. M/5/HK.04.00/V/2025), recruitment fee scams, Telegram/WhatsApp task-deposit fraud, sub-UMP daily/monthly wage offers, and unverified recruiter identities, with instant text highlighting and PNG report card export. - 02 — Eksperimen 30 Hari AI Lamar Kerja (
/eksperimen.html): Privacy-redacted public telemetry dashboard aggregating 70 real fresh-graduate engineering applications across 54 companies, funnel conversion rates, portal breakdowns (KarirJakarta, Glints, JobStreet, direct portals), and human-in-the-loop quality lessons. - 03 — Simulator Sidang Skripsi (
/sidang.html): Extracts quantitative claims, sensors/controllers, and methodological gaps from an undergraduate thesis abstract, then runs a 3-examiner oral defense simulation (Prof. Bambang on methodology, Dr. Ratna on industry edge cases, Pak Hendra on statistical validity) with Web Speech API voice synthesis and real-time rubric scoring. - 04 — Peta Gaji Fresh Grad RI vs UMK (
/gaji.html): Compares fresh-graduate salary offers against 2025/2026 city minimum wages (UMK) across 16 Indonesian industrial regions and 5 engineering/tech roles, estimates disposable savings after kos + living costs, and generates polite HR negotiation scripts. - 05 — Roast CV ala HRD (
/roast-cv.html): Client-side CV bullet auditor checking quantitative impact density, active vs. passive Indonesian/English verbs, buzzword cliches, and PII leaks (16-digit NIK), paired with 3 reviewer personas (HRD Startup Jaksel, HRD Manufaktur, Mesin ATS) and an automatic Google-XYZ bullet rewriter. - 06 — Datasheet Register → C Driver Generator (
/datasheet.html): Parses Markdown/TSV I2C & SPI register maps (including built-inINA219,MPU-6050, andBMP280presets) into production-style.hbitmask headers,.cdrivers for STM32 HAL (HAL_I2C_Mem_Read/Write) or ESP-IDF, and deterministic loopback unit tests. - 07 — Pasal & Kontrak Diff Explainer (
/pasal.html): Longest-Common-Subsequence (LCS) word-level diff engine comparing UU Ketenagakerjaan vs. UU Cipta Kerja provisions and standard vs. predatory PKWT employment contracts, translating legal wording changes into concrete worker-rights impacts.
What was checked
vanovri/radar-loker scanning a predatory job posting (score 100/100) alongside the 7-tool navigation bar.Release audit: 25 October 2026. All 7 client-side engines pass 15 / 15 automated Node.js unit tests (tests/rules.test.mjs and tests/suite.test.mjs) and headless Chromium Playwright checks across all 7 pages with zero console errors.
What this demonstrates
Privacy-preserving client-side text analysis, deterministic rule engines, embedded C code generation from hardware register specifications, and word-level legal diffing without server-side dependencies or external LLM API keys. Legal and regulatory references are educational summaries, not formal legal counsel.