I applied online. I interviewed at Lyft in Jan 2022
Interview
Applied from LinkedIn, and reached out by a recruiter in two weeks. The recruiter is very nice and helpful.
Scheduled the video call two weeks later and go over a case study in 45 mins.
Took 10 mins to introduce each other and related experiences.
Then the interviewer sent a link to show the question and data attributes.
The question is very vague, and you need to define something by yourself.
It wasted a lot time to clarify things.
If you are familiar with their business, the question is simple, but need some time to think and clarify.
The interviewer (a senior data scientist) is kinda arrogant and if you cannot answer the questions quickly then he will give you the answer directly.
I also can feel although the interviewer is senior but lack of knowledge about state-of-the-art machine learning models, like representation learning and latent factor models.
Interview questions [1]
Question 1
Predict a passenger will be churned or not, given some transaction data.
Like a simple Kaggle data science competition, but much vague and no clear definition.
Advantages and limitations of two classifiers.
I interviewed with a recruiter last fall. Since I was graduating in eight months, she said she'd talk to the hiring manager about whether to send technical interviews now or hold off for different openings in the future, and would get back to me in a few days. She never followed up and didn't respond to my emails. Eventually, I got an automated rejection, the subject line literally read "Update on [insert job title]," they hadn't even bothered to fill in the job title. The whole process felt unprofessional.
“I completed all interview rounds — including HR, technical screening, product sense, business case, algorithm live coding, decisions live coding, machine learning, and experience interviews — but the company ultimately selected another candidate
Interview questions [1]
Question 1
Have a good understanding of stats/probabilities/ML/Coding(SQL,Python), and business domain and metrics.