Machine Learning Engineer · Phone Screen · Coding

Microsoft — Machine Learning Engineer ❌ Failed

Level: Senior-Level

Round: Phone Screen · Type: Coding · Difficulty: 6/10 · Duration: 45 min · Interviewer: Very Unfriendly

Topics: Machine Learning, Model Selection, XGBoost, LLM, Dataset Quality, Hugging Face

Location: Seattle, WA, US

Interview date: 2026-01-05

Got offer: False

Summary

I had a phone screen interview for a Machine Learning Engineer position. After introductions, I was asked about classic ML basics like model selection and the pros and cons of XGBoost. Then, I had a whiteboard coding question about checking the quality of an LLM training dataset.

Details

The interview started with a 10-minute introduction, followed by 15 minutes of high-level discussion on classic ML basics, such as how to select a model based on the problem and the advantages and disadvantages of XGBoost. For the last 15 minutes, I received a whiteboard coding question: write code to perform a quality check on an LLM training dataset. The dataset consists of prompt + Excel table + response. I also had to use the dataset to test whether a Hugging Face pre-trained model needed fine-tuning to complete these tasks. In the last 5 minutes, I asked a question, which the interviewer answered briefly, and the interview ended two minutes early. The interviewer was not very conversational and seemed uninterested.