Machine Learning / AI
How did you implement Text-to-SQL? How did you measure 91% accuracy?
Which tools and tech stack have you worked with the most?
What is ring-allreduce and why is it more efficient than a parameter server?
What are the options for sampling new tokens in LLM?
```python import asyncio import time async def work(): await asyncio.sleep(1) async def main(): start = time.perf_counter() await asyncio.gather(*(work() for _ in range(5))) print("Total time:", time.perf_counter() - start) asyncio.run(main()) ``` ```python import time def work(): time.sleep(1) start = time.perf_counter() for _ in range(5): work() print("Total time:", time.perf_counter() - start) ```
How is the pipeline for binary classification with a large number of features and data organized?
Tell about ensembles in machine learning.
How does a tree handle feature emissions that are missing in the training set?
How do you generally gather requirements? Was this task presented as an idea or a detailed assignment?
How would you implement protection against offensive/provocative questions so that the bot does not start swearing in response?
Tell me about Momentum: what are its features?
Is there experience in improving inference or training frameworks for LLM?
How does the K-means clustering algorithm work? Please answer this question orally.
How to split data into groups using window functions in SQL?
Tell us about your practical experience with A/B testing
Can you explain how the F1-score is calculated using precision and recall?
How are collisions resolved in a set? What collision resolution methods exist?
How does contrastive loss with in-batch negatives work?
Tell us about your career overview: roles, industries, key projects.
What is RegNet and the approach of the design space?