## Round 1 - Coding ## Problem Implement a mini ML toolkit without NumPy or sklearn. You will be asked to code one or more of the following from scratch during the interview. **Part A:** Implement logistic regression with gradient descent. ```python def sigmoid(z: float) -> float: ... def train_logistic_regression( X: list[list[float]], y: list[int], lr: float = 0.01, epochs: int = 100 ) -> list[float]: # returns weights ... def predict(X: list[list[float]], weights: list[float]) -> list[int]: ... ``` **Part B:** Implement k-means clustering. ```python def kmeans( points: list[list[float]], k: int, max_iter: int = 100 ) -> list[int]: # cluster label per point ... ``` ``` Example (Part A): X = [[1, 2], [3, 4], [5, 1]] y = [0, 1, 1] weights = train_logistic_regression(X, y, lr=0.1, epochs=200) predict([[2, 3]], weights) -> [1] ``` ## Follow-ups 1. How do you detect if gradient descent is diverging? 2. What initialization strategy would you use for k-means to avoid bad local optima? 3. How would you vectorize your logistic regression using only lists? 4. How do you evaluate model quality without sklearn metrics?

InterviewDB Experience

ML Coding Round - Implement Common ML Primitives from Scratch

Onsite

Interview Experience

Round 1 - Coding Problem Implement a mini ML toolkit without NumPy or sklearn. You will be asked to code one or more of the following from scratch during the interview. Part A: Implement logistic regression with gradient descent. Part B: Implement k-means clustering. Follow-ups How do you detect if gradient descent is diverging? What initialization strategy would you use for k-means to avoid bad local optima? How would you vectorize your logistic regression using only lists? How do you evaluate…

Full Details

🔒

Unlock all Scale AI questions

Full insider details, leaked discussions, and candidate experiences.

Get full access — $100 a year, unlimited access

About This Question

This is a candidate experience report from a scale ai interview during the onsite round.

It covers the following topics: Coding, Mle, Onsite .

Topics

Coding Mle Onsite

ML Coding Round - Implement Common ML Primitives from Scratch

Interview Experience

Full Details

About This Question

Topics

More Scale AI Interview Questions