Design Bot Detection - System Design | WinJob

Problem Statement & Context

Design a global edge bot detection platform (Cloudflare Bot Management / PerimeterX class) that stops automated abuse before origin while preserving crawlers and humans. This section focuses on mission at the CDN edge before origin sees abusive automation.

1.1: edge-first scoring
1.2: good-bot allowlists
1.3: CAPTCHA escalation

Mechanism

Every HTTP request passes through a bot score pipeline that fuses TLS fingerprints, JS telemetry, and rate signals before WAF or origin enforcement.

Interview phrasing

Lead with 12M RPS edge evaluations, 3ms p99 decision budget, <0.1% human false-block, and 99.99% scorer availability.

Failure mode to volunteer

During credential-stuffing surge, Redis shard hot-spots; you micro-shard ASN counters and widen challenge band instead of hard-blocking checkout.

Signal	Target
Peak RPS	12M
p99	3 ms
False block	0.1 %
Availability	99.99 %

javaOne Dark Pro

1 public record EdgeRequest(String requestId, String ja3Hash, String ip, String asn) {}

pythonOne Dark Pro

1 def bot_band(score: float) -> str:
2     if score >= 0.92: return "block"
3     if score >= 0.65: return "challenge"
4     return "allow"

typescriptOne Dark Pro

1 export type BotAction = "allow" | "challenge" | "block";
2 export interface EdgeScore { score: number; action: BotAction; reasons: string[]; }

Why interviewers care

Bot Detection interviews reward crisp scope, explicit trade-offs, and failure stories—not generic microservice diagrams.

Interview checkpoint

Name one failure story for Problem Statement & Context that proves you understand real outages, not happy-path diagrams.

Key Highlights

•Problem Statement & Context: mission at the CDN edge before origin sees abusive automation
•Mechanism: edge-first scoring
•Metric anchor: 12M RPS / 3 ms

1	def bot_band(score: float) -> str:
2	if score >= 0.92: return "block"
3	if score >= 0.65: return "challenge"
4	return "allow"

1	export type BotAction = "allow" \| "challenge" \| "block";
2	export interface EdgeScore { score: number; action: BotAction; reasons: string[]; }