I applied through an employee referral. The process took 4 months. I interviewed at Quantexa
Interview
I first met the Principal Scientist. This gave me an overwhelmingly positive impression that the team is well managed.
Then I had a culture fit interview with the talent partner which was a delight.
This was followed by a technical assessment. For this, I was handed a project with clear instructions and a clear goal. The instructions included some constraints in its scope and technologies to be used which made it interesting and engaging. What they value the most is your approach and reasoning rather than raw scores which also gave me a positive impression and they give you more than enough time to imporove your model's performance and refine your deliverables.
Next step was an interview with two people from the team. This interview is just a chat and not a project assessment. The people I met with were nice and smart. They asked me questions about my background and some technical questions regarding my skills.
The final step was a compound interview which included two phases. The review of the technical project and a whiteboard assessment. This interview was carried out by the Principal Scientist and another colleague from the team. I enjoyed my time presenting the my project. The discussion we had around the project was interesting and engaging. The whiteboard assessment part of the interview was carried out like a brainstorming session. I throughly enjoyed designing a solution for the problem they had in mind for this assessment. This was followed by and interesting discussion as well.
Overall, the interview process is not easy and short. However, this is a good indication that if you get accepted, you will be working with people with the same mindset and similar approaches to problems as you do. I believe this is what takes a team from good to great. Even if I didn't get accepted, I would still be happy to have gone through it because this was also a learning process.
Interview questions [1]
Question 1
How would you design a system to cluster high-volume streaming data where scalability is a priority?