Rate limiting and abuse prevention in Prompt Engineering / GenAI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding Rate Limiting Purpose

Why is rate limiting important in AI service APIs?

A To increase the speed of AI model training

B To prevent excessive use that can overload servers and degrade service quality

C To improve the accuracy of AI predictions by limiting data input

D To reduce the size of AI models for faster downloads

💡 Hint

Think about what happens when too many requests come at once.

Rate limiting helps protect servers from too many requests at once, which can cause slowdowns or crashes.

❓ Model Choice

intermediate

Choosing a Rate Limiting Strategy

Which rate limiting strategy allows a user to make bursts of requests but limits the average rate over time?

A Fixed window limiting

B No rate limiting

C Token bucket algorithm

D Leaky bucket algorithm

💡 Hint

This algorithm uses tokens to allow bursts but controls the overall rate.

The token bucket algorithm lets users send bursts of requests if tokens are available but enforces an average rate limit.

❓ Predict Output

advanced

Output of Rate Limiting Code Snippet

What is the output of this Python code simulating a fixed window rate limiter?

Prompt Engineering / GenAI

import time

class FixedWindowLimiter:
    def __init__(self, limit, window_seconds):
        self.limit = limit
        self.window_seconds = window_seconds
        self.request_times = []

    def allow_request(self):
        current_time = time.time()
        # Remove requests outside the current window
        self.request_times = [t for t in self.request_times if t > current_time - self.window_seconds]
        if len(self.request_times) < self.limit:
            self.request_times.append(current_time)
            return True
        else:
            return False

limiter = FixedWindowLimiter(limit=3, window_seconds=5)
results = []
for _ in range(5):
    results.append(limiter.allow_request())
    time.sleep(1)
print(results)

A [True, True, False, False, False]

B [False, False, False, False, False]

C [True, True, True, True, False]

D [True, True, True, False, False]

💡 Hint

The limiter allows 3 requests per 5 seconds window.

The limiter allows the first 3 requests immediately, then blocks the next 2 because they fall within the same 5-second window.

❓ Metrics

advanced

Evaluating Abuse Prevention Effectiveness

Which metric best measures how well a rate limiter prevents abuse without blocking legitimate users?

A False positive rate (blocking good users)

B Average response time of the server

C Number of API endpoints

D Model accuracy on training data

💡 Hint

Think about mistakenly blocking users who should be allowed.

False positive rate shows how often legitimate users are wrongly blocked, which is key for abuse prevention quality.

🔧 Debug

expert

Debugging Token Bucket Implementation

Given this token bucket rate limiter code, what error will occur when calling allow_request() repeatedly without waiting?

Prompt Engineering / GenAI

import time

class TokenBucket:
    def __init__(self, capacity, refill_rate):
        self.capacity = capacity
        self.tokens = capacity
        self.refill_rate = refill_rate
        self.last_refill = 0

    def allow_request(self):
        current_time = time.time()
        elapsed = current_time - self.last_refill
        self.tokens = min(self.capacity, self.tokens + elapsed * self.refill_rate)
        self.last_refill = current_time
        if self.tokens >= 1:
            self.tokens -= 1
            return True
        else:
            return False

bucket = TokenBucket(5, 1)
for _ in range(10):
    print(bucket.allow_request())

A NameError because time module is not imported

B ValueError due to negative tokens

C No error, prints True for first 5 calls then False

D ZeroDivisionError in refill calculation

💡 Hint

Check if all required modules are imported.

The code uses time.time() but does not import the time module, causing a NameError.