25 high-frequency technical interview questions asked at top Indian analytics firms and MNCs: bias-variance tradeoff, SQL window functions, cross-validation, and live coding challenges with solutions.
Technical interview rounds for Data Scientist and Machine Learning roles at firms like Amazon India, Accenture, Deloitte, Swiggy, and Tier-1 GCCs evaluate theoretical depth, SQL fluency, and live algorithmic coding. Below are high-frequency questions with complete technical explanations.
Bias is error introduced by approximating a complex real-world problem with an overly simplistic model (leading to underfitting).
Variance is error introduced by extreme sensitivity to small fluctuations in training data (leading to overfitting).
Bagging (Random Forest) reduces variance by averaging predictions of uncorrelated deep trees. Boosting (XGBoost/LightGBM) sequentially corrects errors of weak shallow learners, reducing both bias and variance.
Candidates are frequently asked to compute 30-day rolling sums and employee rankings within departments using SQL window functions:
-- SQL Window Functions Asked in Tier-1 Tech Interviews
SELECT
employee_id,
department,
salary,
-- 1. Consecutive rank without gaps
DENSE_RANK() OVER (PARTITION BY department ORDER BY salary DESC) as salary_rank,
-- 2. Running cumulative payroll by department
SUM(salary) OVER (
PARTITION BY department
ORDER BY employee_id
ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW
) as cumulative_payroll
FROM employees;Standard K-Fold randomly splits data into K subsets. On severely imbalanced datasets (e.g. 98% non-fraud, 2% fraud), a random fold might contain zero positive fraud cases, severely destabilizing evaluation.
Stratified K-Fold guarantees that each fold preserves the exact class percentage ratio of the complete dataset, ensuring reliable evaluation metrics.
Datadrix conducts weekly 1:1 mock technical interviews with senior industry architects to guarantee you walk into your placement rounds with total confidence.
Join Datadrix’s live cohort with 1:1 mentor code reviews, capstone projects, and 400+ hiring partner referrals.
A practical guide for students and professionals in Ghaziabad, Mohan Nagar, Raj Nagar, and Vaishali looking to master Python: fundamental concepts, project expectations, and how Datadrix’s Dilshad Colony campus bridges East Delhi and Ghaziabad.
A complete buyer's guide and 10-point evaluation checklist for choosing the best data science course in Delhi NCR: syllabus depth, mentor pedigree, live capstone projects, placement assistance, and avoiding vanity certificates.
Take machine learning from Jupyter notebook to the cloud: data pipelines, model training with Scikit-Learn, async REST serving with FastAPI, Docker containerization, and AWS EC2 deployment.