What this generates
Realistic-looking fake user records with these fields per user:
- UUID (real crypto-secure UUIDv4)
- First and last name
- Email (formatted from the name)
- Username (firstname+lastname+digits)
- Age (18-78)
- Role (Software Engineer, Product Manager, etc.)
- Company (Acme Corp, Globex, etc., fake businesses)
- City (12 major world cities)
- Phone (formatted international)
- Avatar URL (DiceBear identicon based on username)
Pick how many you want (1-100) and the output format (JSON, CSV, or SQL INSERT). The card preview shows the first six users with avatars; the full data appears below as text you can copy.
When you’d use this
- Database seeding: bootstrap a dev or test environment with users that don’t reveal real data
- UI mockups: Figma or HTML prototypes that need filled-in placeholder content
- Demo videos: real-looking but obviously-fake users for product demos
- Test fixtures: integration tests need consistent fake data structures
- API testing: webhook payloads with realistic shapes
Names are diverse on purpose
The first-name and last-name lists pull from many cultures, Latin, Slavic, Arabic, Indian, Chinese, Japanese, African, European. You’ll get realistic-looking combinations from any region. Some pairings are linguistically odd (“Naomi Nakamura” is fine, but “Sami Schmidt” is unusual), that’s a feature, not a bug. Real datasets always have surprising name combinations.
Output formats
JSON: standard pretty-printed JSON array. Drop into a fixtures file or seed script.
CSV: comma-separated with headers. Quote-escaped for fields containing commas or quotes. Imports cleanly into spreadsheets, databases, and ETL pipelines.
SQL: ready-to-run INSERT INTO users (...) VALUES (...) statements with single-quoted strings and proper escaping. Drop into a .sql migration file.
What this isn’t
- Not a real user generator: every record is fake. Don’t use these emails to send actual mail; the addresses follow a real format but go nowhere.
- Not GDPR-equivalent test data: real personal data has subtle distributions (age skews, name frequencies, ZIP code clusters) that fake data doesn’t replicate. For ML training data, use representative real-world samples.
- Not for stress testing at scale: 100 users is the cap. For 10K+ row generation, use a CLI tool like Faker (Python) or Mockaroo.
Frequently asked questions
Are the avatars real photos? No. They’re SVG identicons generated from the username via DiceBear. Free for any use, no attribution needed.
Can I customize the field set? Not yet. The fields are fixed. For custom schemas, use Mockaroo or a code-based generator (Faker.js, Faker.py).
Are the UUIDs real?
Yes, generated via crypto.randomUUID(), RFC 4122 UUIDv4 format. Statistically unique.
Why are the same names repeating in big batches? The name lists are about 70 first names and 50 last names. With 100 users, you’ll see some repeats. Real datasets do this too. For uniqueness guarantees, deduplicate after generation.