The second round was with a senior engineer from the same team. it was a coding interview on CoderPad(the research team uses mostly python). 2 questions, first one is basically the implementation of the equivalent of scipy.stats.rv_discrete.
Applied Research Interview Questions
145 applied research interview questions shared by candidates
how to get all combination from 4 list, each list contains certain numbers.
Artificial Intelligence and Deep Learning, Reinforcement learning.
The test was about floating-point arithmetic; in the first task, I had to implement a very simple floating-point operation considering possible optimizations. In the second task, I had to compare and choose the best between two given implementations of a simple mathematical operation and justify the answer.
Lot of things asked during the paperwork, documentation needed to be provided for obvious reasons. Extremely thorough.
1. Improvements upon the classical attention (in LLMs) that can save compute/arithmetic operations.
Are you familiar with the BART model
List some model parallelism paradigms that can tackle the low throughput issue in model training/serving
NDA, but very simple questions. For non NDA question, describe your related work experience.
What are the loss function components of single stage detectors?
Viewing 1 - 10 interview questions
See Interview Questions for Similar Jobs
Applied ScientistDeep Learning ScientistApplied Research ScientistDeep Learning Research EngineerDeep Learning EngineerDeep Learning / Machine Learning ResearcherArtificial Intelligence EngineerJunior Machine Learning EngineerMl EngineerJr Data ScientistApplied Machine Learning EngineerQuantitative Research InternshipDeep Learning Data ScientistApplied Machine Learning ResearcherAi Machine Learning EngineerArtificial Intelligence ConsultantMachine Learning ResearcherArtificial Intelligence Developer