Math Questions at ByteDance: What to Expect

Math questions appear in roughly 12% of ByteDance’s technical interviews (8 out of 64 total problems). They aren't just academic exercises. ByteDance’s core products—TikTok, Douyin, Toutiao—rely heavily on algorithms for content recommendation, real-time analytics, ad pricing, and massive-scale data processing. A strong math foundation directly translates to optimizing these systems, whether you're calculating probabilities for A/B tests, designing efficient data structures, or modeling user engagement. Interviewers use these problems to assess your analytical rigor and your ability to translate a real-world problem into a precise, computable model.

What to Expect — Types of Problems

ByteDance’s math questions typically fall into three categories. Probability and Statistics are most common, involving scenarios like card draws, game strategies, or estimating metrics from sampled data. Combinatorics problems ask you to count arrangements or selections under constraints, often requiring careful case analysis. Finally, Number Theory and Modular Arithmetic appear in problems related to hashing, cryptography, or optimizing operations with large integers. The problems are often presented as word problems or embedded within a coding task, testing both your mathematical modeling and your ability to implement the solution efficiently.

How to Prepare — Study Tips with One Code Example

Focus on core concepts: probability rules (conditional, Bayes’), combinatorics formulas (permutations, combinations), and modular arithmetic properties. Practice translating word problems into equations before coding. A key pattern is using dynamic programming (DP) for probability and combinatorics where outcomes depend on previous states.

For example, consider finding the probability of reaching a point on a number line starting from zero, moving +1 with probability p and -1 with probability 1-p in n steps. This is solved with DP where dp[i][j] represents the probability of being at position j after i steps.

def probability_after_steps(n, p, target):
    # dp[step][position] -> probability
    # Position range: -n to n, offset by n for index
    dp = [[0.0] * (2 * n + 1) for _ in range(n + 1)]
    dp[0][n] = 1.0  # Start at position 0 (index n)

    for step in range(1, n + 1):
        for pos in range(-n, n + 1):
            idx = pos + n
            # Came from left
            if pos - 1 >= -n:
                dp[step][idx] += dp[step - 1][idx - 1] * p
            # Came from right
            if pos + 1 <= n:
                dp[step][idx] += dp[step - 1][idx + 1] * (1 - p)

    return dp[n][target + n]

Recommended Practice Order

1. Master Fundamentals: Review probability, combinatorics, and modular arithmetic. Solve textbook-style problems manually.
2. Apply to Code: Practice implementing the above concepts in your primary language using platforms like LeetCode. Filter for ByteDance math-tagged problems.
3. Simulate Interviews: Time yourself solving ByteDance’s past math questions. Explain your reasoning aloud as you derive formulas and write code.
4. Identify Weaknesses: If you miss a problem, categorize the error—was it the mathematical model or the implementation? Drill that category.

Practice Math at ByteDance