Episode 69
How To Crack Machine Learning Interviews (Microsoft & Walmart Sr Data Scientists POV) - w/ Nirmal & Karun
One of the twenty most-watched Ready Set Do episodes on YouTube right now.
Machine learning interviews have become a strange mix of theory, product sense, and please-do-not-waste-my-time energy. Nirmal and Karun pull the curtain back on what candidates keep getting wrong, what hiring teams actually notice, and how to stop rehearsing answers that sound smart but do not land.
Machine learning interview prep from Microsoft and Walmart senior data scientists.
This episode is the Ready Set Do machine learning interview playbook with Nirmal Budhathoki, introduced as a senior data and applied scientist at Microsoft, and Karun Thankachan, introduced as a senior data scientist at Walmart. The transcript covers how data scientist, machine learning engineer, applied scientist, and AI engineer loops differ; what coding, SQL, ML theory, statistics, deep learning, MLOps, system design, project discussion, and behavioral rounds test; and how candidates should prepare without drowning in generic ML resources.
- Nirmal Budhathoki is introduced in the episode as a senior data and applied scientist at Microsoft.
- Karun Thankachan is introduced in the episode as a senior data scientist at Walmart.
- The transcript distinguishes data scientist, machine learning engineer, applied scientist, and AI engineer interview loops.
- Karun says his first Amazon interview opened with linear regression assumptions, and he fumbled because his projects had been focused on language models.
- Nirmal uses class imbalance beyond SMOTE as a favorite way to test whether candidates understand real modeling tradeoffs.
- The conversation discusses Decoding Machine Learning Interviews as a curated 100-question fundamentals resource for candidates with limited prep time.
Who this is for
- You are trying to get hired without sounding like everybody else in the pile.
- You would rather hear Nirmal's version while the mess is still fresh than get another polished hindsight sermon.
Key takeaways
- Crack Machine Learning Interviews (Microsoft & Walmart Sr Data Scientists POV) - w/ Nirmal & Karun
- Building a Strong ML Foundation as a Student
- Nirmal and Karun pull the curtain back on what candidates keep getting wrong, what hiring teams actually notice, and how to stop rehearsing answers that sound smart but do not land.
- If you are trying to get through the ML interview maze without losing a week to bad advice, this one is worth it.
- Nirmal says the first mistake is trusting the job title too much. Data scientist, machine learning engineer, applied...
- Nirmal's practical advice is to reverse-engineer the responsibilities instead of reacting to the title. If the job...
Need the cleaner version?
I pulled the sharpest parts of this lane into a guide so you do not have to reconstruct the answer from memory later.
Fast scan timestamps
Read the complete transcript
The edited transcript keeps the speakers' ideas, timestamps, and specific details while removing repetition and filler.
Transcript-backed moments
A few lines worth stealing before you hand over the full hour.
Naman opens the episode with the problem most machine learning candidates run into: interviews do not only test whether someone has touched models. They test whether the candidate can explain fundamentals, code under pressure, discuss projects, and handle follow-up questions without hiding behind buzzwords. He introduces Nirmal Budhathoki, a senior data and applied scientist at Microsoft, and Karun Thankachan, a senior data scientist at Walmart, as the guests for a practical conversation on cracking machine learning interviews.
Naman also sets up their book, Decoding Machine Learning Interviews, as the backbone for the conversation. The book is framed as a focused resource with 100 curated machine learning interview questions and answers. The episode stays practical: ML interview prep, data scientist and machine learning engineer loops, applied scientist screens, and the technical expectations candidates encounter across large technology companies.
Naman starts with the structure of a machine learning interview. He asks what rounds show up, what kinds of questions candidates should expect, and how someone should prepare before jumping into individual machine learning concepts.
Nirmal says the first mistake is trusting the job title too much. Data scientist, machine learning engineer, applied scientist, and AI engineer can mean different things across companies. A candidate has to read the job description and understand whether the role is core machine learning, production ML engineering, applied science, analytics, or GenAI. For core ML data science roles, he expects strong statistics, machine learning foundations, deep learning basics, and enough awareness of current GenAI trends without pretending every role is a GenAI research job.
Nirmal's practical advice is to reverse-engineer the responsibilities instead of reacting to the title. If the job description talks about building ML models, evaluating models, applying statistics, and understanding foundational machine learning and deep learning, the candidate should prepare for a core ML-focused data science loop. If the role is explicitly AI engineer or GenAI-heavy, the prep has to tilt more toward generative AI systems, but that is not the default for every machine learning role.
Show notes
Machine learning interviews have become a strange mix of theory, product sense, and please-do-not-waste-my-time energy. Nirmal and Karun pull the curtain back on what candidates keep getting wrong, what hiring teams actually notice, and how to stop rehearsing answers that sound smart but do not land. If you are trying to get through the ML interview maze without losing a week to bad advice, this one is worth it.
FAQ
How do you prepare for machine learning interviews?
Nirmal and Karun recommend splitting prep by role: coding and DSA for screens, SQL or Python where relevant, ML theory and statistics, project discussion, case or MLOps questions, ML system design for more senior loops, and behavioral practice. The transcript is useful because it explains which parts matter for data scientist, machine learning engineer, applied scientist, and AI engineer roles.
How much LeetCode do machine learning engineer candidates need?
Nirmal says it depends on how engineering-heavy the role is. A production machine learning engineer role at a top tech company can have a coding bar close to software engineering, while many data scientist or applied scientist roles can be covered with medium-level coding plus stronger ML, statistics, and business reasoning.
What projects help master's students get ML internships?
Karun recommends starting with guided projects to learn the ML lifecycle, then moving into business-relevant Kaggle competitions, especially completed competitions with real prize money. Nirmal adds that strong projects should expose candidates to edge cases, discussion boards, alternative model choices, and a clearer problem formulation.
What linear regression questions come up in ML interviews?
Karun highlights linear regression assumptions because that question caught him in an Amazon interview. Nirmal extends it into residual plots, R-squared, adjusted R-squared, and the trap where adding features can make R-squared look better without meaningfully improving the model.
What should ML candidates know about PCA in interviews?
Nirmal says candidates should not stop at defining PCA. Stronger answers explain where PCA fails, what its disadvantages are, what alternatives exist, and why every ML concept has a flip side that can become an interview follow-up.
How should candidates answer class imbalance questions beyond SMOTE?
Nirmal says many candidates only answer with SMOTE. He pushes for more depth: SMOTE can amplify noisy minority examples, and candidates should also discuss class weights, heavier penalties for minority-class mistakes, metric choice, and the parameters available in libraries like scikit-learn.
Can ChatGPT help with ML interview prep?
Yes, but both guests warn against generic use. Karun suggests generating and answering topic-specific questions, while Nirmal recommends feeding the model the job description, recruiter notes, role responsibilities, and likely edge cases so the practice is specific to the actual interview.
Why use a 100-question ML interview book instead of a huge question bank?
Nirmal says the internet already has too much ML information. Their goal with Decoding Machine Learning Interviews is a concise set of repeated public interview patterns and fundamentals that candidates can realistically review in a short preparation window.
More in Get Hired
Same mess. Different guest. Pick the next conversation that feels closest to your real life.