ML theory interview practice
Answer machine learning theory questions out loud: generalization, training deep networks, evaluation, language models and debugging a model that misbehaves. Vera probes each answer once or twice for depth and practical judgment, then moves on.
- 35 min
- Six at regular, five at lead
- At most one small hint
- Machine learning
How the round runs
- 01
The question
Vera asks one question and lets you answer. Start with a short, accurate answer, then say when you would use it.
- 02
Probes for depth
One or two follow-ups on the same idea: why it works, what breaks if a condition fails, when you would not use it.
- 03
The next question
Once there is enough signal, Vera moves on. Vera does not teach during the round; the report says what was missing.
What the report scores
Each skill gets a score from 1 to 5 against the bar for your level, with the quotes it is based on and the time you said them. The report ends with one thing to fix next.
- Accuracy
- Depth
- Practical judgment
- Communication
At lead level, also: Breadth and connection.
How the real round works
The ML theory round, also called ML breadth or ML fundamentals, is a run of short spoken questions: bias and variance, regularization, losses and optimizers, evaluation metrics, data leakage, training problems in deep networks and, more and more, the basics of language models. The guides we read agree that a textbook definition is only the starting point. The signal is applied judgment: when you would use a method, which knob you would turn for a given symptom, and how you would check that it helped.
How ML theory interviews work: structure, signals and common mistakesRegular and lead level
- Regular
- Accurate answers that explain how a method works, with a real example of when you would use it.
- Lead
- Also breadth and connection: tie each answer to its consequences for data, training, production and the team’s decisions.
Length: 35 minutes at either level.
See the feedback before you start
Full reports from the real report pipeline, for a fictional student and a scripted interview.
Questions
What topics come up?
Core ML such as generalization, regularization and classical models, optimization, evaluation, deep learning architectures, embeddings, language models, recommendation, and the practice of debugging and shipping models.
Is this the same as the ML coding round?
No. This round is spoken questions. ML coding is a separate round where you implement models, metrics and data code in an editor that runs them.
Will Vera tell me the right answer?
No. Vera probes for depth and gives at most one small hint in the round. The report afterwards shows what was strong and what to work on next.
What does the report score?
Accuracy, depth, practical judgment and communication, plus breadth and connection at lead. Each score quotes what you said.