Me postulé en línea. El proceso tomó 2 semanas. Acudí a una entrevista en Tensor Labs (Islamabad, ) en ene 2025
Entrevista
One of the best interview processes I've been through, appeared 2 times for interview and got selected the 2nd time. The reason for appearing the first time was their professionalism and the guidance from the person taking the interview right after the call on what things I can improve and what pathway i should take.
Preguntas de entrevista [1]
Pregunta 1
Majority questions were around tabular machine learning and data understanding.
Me postulé a través de una agencia de empleos. El proceso tomó 1 semana. Acudí a una entrevista en Tensor Labs (Islamabad) en oct 2025
Entrevista
There are a total of 2 technical rounds, and the 2nd interview is conducted by the CEO himself, and it is technically challenging. Interview is online and consists of 40 minutes.
Preguntas de entrevista [1]
Pregunta 1
Use of Dropout while training a model and once trained, how to use it during test process.
Complete ML workflow process:
The techniques used for Preprocessing a dataset
Which model to use (based on Classification, Regression etc. problem)
Training, Validation and Testing
Deployment
Search for questions that can be asked in similar scenario
Classification/Regression algorithms in detail:
Which algo to use in a specific use case and why?
Transformer architecture
Bias, Variance, Overfitting, Underfitting:
Ways to overcome Under and Over fitting
Scenario: If a Model shows a graph with a lot of variations (up and down), it refers more
towards underfitting as model is unable to understand the comple and data and is making
assumptions.
Part 2
Good Python concepts:
Typecasting
args and kwargs
yield
FastApi
Database:
How to handle large datasets.
User provides their data and how you will create a profile from this.
Profiling means # of columns/rows, empty values, other stats about the data
