Browser-First Architecture
How MiniJudge Handles Your Data
You are analyzing business datasets. Data confidentiality is not an afterthought; it is our primary architectural constraint.
The Golden Rule
Your source CSV never leaves your browser by default. When you drop a 50,000-row file into MiniJudge, the file is parsed inside your local browser memory via Papa Parse.
1. What stays local on your device
- Your entire source spreadsheet file and unselected columns.
- Columns containing sensitive IDs, phone numbers, or proprietary notes not referenced by your criteria.
- Full merged export output file generated directly in memory.
2. What is sent to the MiniJudge API
- During Step 2 (Describe): Column headers and a small anonymized slice of 8–10 rows are sent to compile your natural language criteria into a structured specification.
- During Step 4 (Run): Only the rows currently being evaluated in active batches (chunks of 50 rows) are sent over TLS 1.3 encrypted HTTPS channels to determine match tiers.
3. Zero Permanent Storage in V0
MiniJudge does not store your rows in databases. We do not train models on your uploaded data. Once a judgement chunk is classified and returned to your browser, it is evicted from server memory.
4. Payment Processing
When unlocking files for €4.90, payment details are processed directly by Stripe via PCI-DSS Level 1 compliant infrastructure. MiniJudge never sees or stores your credit card number.