Yandex Machine Learning Engineer Interview Questions
& Process
Real candidates share what happened, how many rounds they had,
and how the experience turned out.
Based on 21 interview experiences · FREE TO READ
Candidate interview experiences
First-hand accounts from people who interviewed at Yandex.
Yandex
Machine Learning Engineer
Mid Average Positive experience Decline offer 4 rounds
Interview process
Technical screen Onsite
Interview formats
Coding Technical System Design
It was a 4-round process. First 3 rounds were general interviews covering leetcode, ML theory, and recsys topics. Then, there was one interview per product team. The leetcode questions weren't too difficult, although the final one was a bit tricky, involving geometry or general math rather than a standard data structure problem. I enjoyed that one.
Confirmed questions1 question
- What are some ways to avoid biases when designing a recommendation system?
Yandex
Machine Learning Engineer
Senior Difficult Positive experience No offer 3 rounds
Interview process
Technical screen
Interview formats
Technical Coding
The interview was split into 3 parts: one focused on classical ML and programming, another on algorithms and computer vision specialization. The ML part was quite relevant. Algorithms are a big part of the company culture, so that was fine. However, I was really disappointed with the computer vision section; it barely related to real-world applications.
Confirmed questions3 questions
- Given the code for a Deep Learning model and its training procedure, your task is to correct it.
- Can you provide the mathematical formula for cross-entropy loss?
- Explain the reason behind using an odd-numbered kernel size in convolutional layers.
Yandex
Machine Learning Engineer Intern
Intern Difficult Neutral experience No offer 1 round
Interview process
Technical screen Recruiter call
Interview formats
Coding Technical
This was my first interview experience at Yandex, happening over a year ago. It wasn't the most encouraging process, but I sort of expected it due to how Yandex interviews are generally known to be. They immediately gave me a medium LeetCode problem, which I struggled with a bit. I made some coding errors, but the interviewer decided to ask me more about ML, perhaps to balance things out. Unfortunately, I wasn't prepared for the ML theory part, and I couldn't answer some questions, which didn't look good. As expected, I didn't advance, and the recruiter shared feedback the next day.
Confirmed questions2 questions
- Interval List Intersections problem from LeetCode
- Questions on classical Machine Learning theory
Yandex Machine Learning Engineer Interview Questions
Quoted word for word from Yandex interview reports.
“Find a substring in text T that is equal to string S up to permutation.”
Read reports →“Explain the reason behind using an odd-numbered kernel size in convolutional layers.”
Read reports →“What are the drawbacks of the accuracy metric?”
Read reports →“Regarding the optimization of logistic regression weights, can the gradient descent method become trapped in a local minimum?”
Read reports →“What loss function is employed in the logistic regression algorithm?”
Read report →“Can you provide the mathematical formula for cross-entropy loss?”
Read report →“Can you write the minimum distance between two vectors, v1 = [23, 0, 10] and v2 = [17, -4, 22], such that Func(v1, v2) returns 1 (v1[0] - v2[2])?”
Read report →“What are the regularization methods for Deep Learning Models?”
Read report →“What is gradient boosting, and what would occur if we removed one estimator from it?”
Read report →Formats, difficulty and experience
Across all 21 Yandex interview reports.