JA4 bot detection

Real-time TLS fingerprinting that cut automated payment-API abuse by 45% with zero user friction.

The problem

Bots rotated IPs and user agents faster than blocklists could keep up, hitting payment APIs.

What I built

  • A real-time pipeline computing JA4 and JA4H fingerprints from TLS ClientHellos and HTTP headers
  • Fingerprint reputation scoring fed into WAF decisions at the edge

The result

45% fewer automated abuse attempts on payment APIs, with no added friction for real users.