Netflix Machine Learning Engineer interview questions

15 Machine Learning Engineers have shared notes from their Netflix interviews, with the questions they were asked in each round. That is 8% of the 191 Netflix interview notes on Screna AI.

15 Machine Learning Engineer interview notes

Netflix·Machine Learning Engineer·Technical Phone Screen

Jul 2026
Netflix ML engineer interview that went deep into NLP fundamentals and modern LLM architecture. Pretty technical throughout, no fluff questions, they clearly wanted someone who actually understood the internals of these models and not just how to call an API.
  • What is SentencePiece and how does it work under the hood?
  • Which tokenizers do BERT and typical Transformer-based language models use, and what's the reasoning behind those choices?
  • Walk me through the core components inside a Transformer block and what each one does.
  • Compare a standard Transformer to LLaMA and Qwen architectures. What design choices differ and what are the trade-offs, specifically around things like MoE, RMSNorm, and RoPE?

“I knew the high level answer but stumbled a bit on the specifics of the training algorithm.”

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

Jul 2026
Netflix ML Engineer screen that went pretty deep on optimizer theory. One question, but it had a lot of surface area and they clearly wanted more than a textbook answer.
  • Walk me through the differences between Adam and SGD, including how each update rule works, memory overhead, convergence behavior, and when you'd pick one over the other.

“I started with the mechanics, SGD is just gradient times learning rate, with momentum as an optional velocity term that smooths out the updates.”

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Netflix ML engineer screen, one meaty coding question about parsing model parameter keys into a nested structure. Felt like a mix of pure Python and systems thinking, which I wasn't fully expecting.
  • Given a flat list of model parameter keys like 'layer1.attention.q_proj.weight', write a Python function that groups them into a nested dictionary by layer index, placing components like attention and mlp under the correct layer. Define your grouping logic, handle malformed or duplicate keys, and walk through time and space complexity.

“I started by splitting on dots and pulling the layer prefix out, which felt clean, but then they asked what happens with keys that don't follow the pattern and I kind of stumbled.”

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Interviewed for an ML engineer role at Netflix, just the one question about computer vision experience. Short and hard to read in terms of how it went.
  • Walk me through your background and hands-on experience with computer vision.

“Broad opener but I fumbled it a bit by jumping straight into model architectures before they probably wanted to hear about scope and impact.”

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Interviewed at Netflix for what seemed like an ML or data role involving content classification. The prompt was pretty bare bones so there's not much to go on, but the core task was about categorizing social media posts by topic.
  • How would you build a system to automatically classify social media posts by topic?

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

May 2026
Netflix ML Engineer interview, technical phone screen that went pretty deep into transformer internals. One question but it had a lot of layers to it and I felt like I was being tested on whether I actually understood the math or just memorized the formula.
  • In scaled dot-product attention, why do we divide the dot product of Q and K by the square root of the key dimension before applying softmax?

“I knew the answer existed but fumbled the variance argument at first.”

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

May 2026
ML fundamentals screen for a Netflix MLE role. Two topics back to back: loss functions and LoRA. No coding, just whiteboard reasoning and they pushed hard on the math.
  • Compare MSE and cross-entropy loss. When is each appropriate, what probabilistic assumption does each correspond to, and why is cross-entropy preferred over MSE for classification with a sigmoid or softmax output?
  • Derive the gradient of MSE and cross-entropy with respect to the pre-activation logit for a sigmoid output, and explain precisely why MSE's gradient can stall when the model is confidently wrong.
  • Explain the core mechanism of LoRA: how the weight update is parameterized, what is frozen versus trained, and how the adapter is handled at inference.
  • For a 4096 by 4096 attention projection matrix with rank 8, what fraction of that layer's parameters does LoRA actually train?
  • What do the LoRA hyperparameters rank, alpha, and dropout control, and how would you approach tuning them?
  • How does LoRA compare to other parameter-efficient fine-tuning approaches like adapters, prefix tuning, or QLoRA, and when would you choose one over another?

“I knew the high-level answer but fumbled when they pushed on the likelihood framing.”

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

May 2026
Netflix ML engineer screen, pretty deep on fine-tuning fundamentals. One meaty technical question that took up most of the time, felt like they wanted to see how far you could go with it rather than just checking a definition.
  • Explain LoRA as an approach to fine-tuning large pre-trained models efficiently. Walk through the mechanics, where it gets applied, what you actually gain from it, and what knobs you're tuning when you use it.

“This started straightforward and then got uncomfortable fast.”

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

May 2026
Netflix ML Engineer screen, got a word frequency counter problem which felt more like a warmup than a real challenge. Clean enough problem but I kept second-guessing my design choices out loud which probably didn't help.
  • Design and implement a class that tracks word frequencies from a continuous stream of text, supporting methods to add text, retrieve the count for a specific word, and return all word counts.

“Went with a dictionary under the hood, nothing fancy.”

View Post

Netflix·Machine Learning Engineer·Technical Phone Screen

May 2026
Netflix ML Engineer interview that went deep on Transformer internals. One question but it had a lot of surface area and I probably underestimated how much they'd push on the math.
  • Walk me through how self-attention works in Transformer models, including how Q, K, and V are derived, the scaled dot-product formula, multi-head attention, and why this architecture handles long-range dependencies better than RNNs.

“This is the kind of question where you think you know it until you're actually saying it out loud.”

View Post