The process is a pleasant, 4-steps organized interview. The structure is the following:
1-Form to be filled that is used to adapt thr interview process to your knowledge about the topic;
2- Get to know interview: the guy is really kind and helps you throughout the whole process;
3- Technical interview in which you are asked to solve a simple python programming assignment and to talk about several ML techniques;
4- Practical use-case: You are asked to solve a project and to present in within a week. It should take about 15 hours to be solved, depending on your previous knowledge on the topics. In my case, it has been a pretty intensive work week. Overall the project was funny and taught me a lot.
In conclusion, the guys are really kind, available during all the
process if you need any help. The overall experience is recommended. You will get back the feedback in details of your use-case even when they decide to not move forward with your application.
Interview questions [1]
Question 1
Talk about the reasons why you chose Visium in particular. Why Visium and not any other consulting company?
The process took 4 weeks. I interviewed at Visium in Nov 2023
Interview
The process had 4 steps.
It started with a Human Resources (HR) screening, where basic qualifications and general fit for the organization were assessed.
Then, candidates engaged in a technical chat, delving into their expertise and problem-solving skills in machine learning expertise.
The third phase involved a practical application of skills through a use case assignment, allowing candidates to showcase their hands-on abilities.
Finally, the last step was an interview with the CEO, providing an opportunity for candidates to discuss their vision, alignment with company goals, and overall cultural compatibility.
Interview questions [1]
Question 1
Describe the architecture of a convolutional neural network (CNN)
The process had 3 interviews First interview with HR discussing the role, the company and some additional behavioural questions. The second interview was a technical interview with a Senior ML engineer. He asked several questions on different aspects of ML: anomaly detection, NLP, based tree models, evaluation metrics and stable diffusion. The third interview was a home assignment and a presentation. The main idea is to build and end to end model with a decent accuracy and F1. Be careful with this step of the process since it is time consuming, I had few time since I was working while doing the process so it took a lot of my free time to do it.
Interview questions [1]
Question 1
Could you describe the main based tree algorithms? (RF, XGboost etc...)
