I applied online. The process took 1+ week. I interviewed at Synthesia (New York, NY) in Dec 2022
Interview
Intial call with CTO followed by a take home task for 4 to 6 hours. The task was to design a end to end deep learning model given a data set. At the end, there was a discussion with the team about the code submitted.
Got rejection mail with a standard template after 2 days. Quite disappointing to see template answer even though I invested 8 hours for the interview process.
I applied through a recruiter. The process took 2 weeks. I interviewed at Synthesia (London, England) in Feb 2024
Interview
Approached by their HR, scheduled the first interview with the team lead, Then there's the take home task which were said to take 6 hours to complete, the task is a decoder based language modelling task with an extra dimension. Their HR messed up the time constraint, i was given less than 3.5 hours. Took half an hour reading the task and codes, wasted more than 40 mins trying to set up the local environment. Ended up spending just above 2 hours on it. The feedback was that my solution is quite reasonable, but they have only one opening and obviously the other candidate has been waiting for the offer way before me. It's a challenging task for sure but not impossible if proper time has been given.
Interview questions [1]
Question 1
Decoder based language modelling with extra dimension, modify the codes to make it run and make research decisions in-between.
I applied through a recruiter. The process took 3 weeks. I interviewed at Synthesia (London, England) in Feb 2024
Interview
The interview process was well organised and straightforward. The main technical challenge consisted of a timed take-home assignment with a total duration of 8 hours. Synthesia provided a cloud GPU instance to complete the challenge, which made it feasible to cary out some real ML development, not simply toy tasks.
Interview questions [1]
Question 1
Why did you make this particular decision in implementing specified ML model?
