CoreWeave Technical Phone Interview Data Filtering Algorithm

Completed a technical phone screen for Weights and Biases (now part of CoreWeave). ## Question You're given data like this at Weights and Biases and a common operation is filtering for data you want b

Completed a technical phone screen for Weights and Biases (now part of CoreWeave). ## Question You're given data like this at Weights and Biases and a common operation is filtering for data you want based on your search query. For example: data = [ {'a': 2, 'duration': 20, 'val_acc': 0.9}, {'b': 5, 'loss': 30, 'duration': 25} ] condition = (duration > 22) AND (a > 1 OR b < 3) ### Part 1: Basic filtering Assume that a row of data is a set of key/value pairs where value is float Support < and > comparisons You don't have to worry about string parsing the condition as you can define the representation yourself ### Part 2: Choice of Pick one of: * Print the stringified condition expression based on your part 1 design * Suppport equality condition for strings, booleans, etc I went with part 2 of adding equality condition, but also updated the < or > condition to attempt to convert both the data value and threshold value to float and if we fail, return false ## Reflection * Chill interviewer * Interviewer didn't really engage in the "collaboration part of the design". Rather they gave me free reign over the design and just watched * Asking good clarifying questions to button up the scope helped me get a good handle on the problem