Interview Practice
Technical Screen
Explain, compare, and defend core AI engineering concepts.
What interviewers evaluate
- Precise definitions
- One concrete example
- Stated assumptions and known limits
- Adapting when a constraint changes
What you will be asked
Defend a decision with numbers under uncertainty.
In the room
The interviewer gives measured results for quality, cost, and latency, then asks you to pick an option, state your confidence, and say what evidence is missing.
Timebox
30-45 minutes, 3-5 questions
Mock interview
A timed answer to a random question from this level, then the reference answer and your score.
How to practice
- Work the linked questions on paper first: compute every number before you read the reference answer.
- Practice stating your recommendation in one sentence, followed by the two assumptions it depends on.
- With a peer: they challenge one number (a sample size, a tail latency, or a price change) and you revise the decision.
What a strong answer shows
- Separates mandatory gates from preferences
- Computes the figures correctly and shows the arithmetic
- Treats a result at a threshold as uncertain and asks for intervals or more data
- Gives a provisional decision and what would reverse it
Questions for this level
Each question opens on its Practice tab, and Previous and Next follow this list.
LLM Foundations
4 h
Recommend a model or routing policy from measured quality, cost, and latency, say what evidence is still missing, and revise the decision when the traffic mix changes.
Embeddings & Retrieval
4 h
Design hybrid retrieval with reranking for a multi-tenant help center, choose a configuration from measured quality and latency gates, and keep access rules intact through timeouts and model migrations.
Model Serving & Inference
4 h
Choose how to serve interactive hints and offline drafts on four memory-limited accelerators: compute KV headroom, check measured tail-latency and quality gates, and design scheduling, overload handling, and verification.
AI Product & Experimentation
4 h
Check the validity of a synthetic A/B test of an AI coach, compute its rates, intervals, and costs, and defend a release decision when offline ratings, immediate metrics, and matured outcomes disagree.
Follow-up questions
An interviewer may ask these next.
- Can you average two p95 latencies?
- Your best option meets a gate exactly. Do you ship?
- Traffic shifts toward harder questions. What changes?
- What would you measure in the first week after launch?
Scoring rubric
Interview ready at 8 of 10 with no 0.
| Dimension | 0 | 1 | 2 |
|---|---|---|---|
| Problem framing | Jumps to a solution or solves the wrong problem | Restates the problem but misses key constraints or success criteria | Clarifies goals, constraints, and success criteria before solving |
| Correctness | Key claims or code are wrong | Mostly right with gaps or unhandled cases | Correct and precise, including edge cases |
| Evidence or testing | No way to verify the answer | Generic testing ideas | Concrete fixtures, metrics, or traces that would confirm or refute the approach |
| Trade-offs | One option, no alternatives | Names alternatives superficially | Compares options against the stated constraints and says what would change the decision |
| Communication | Disorganized or unclear | Understandable but meandering | Structured and concise; states assumptions; checks in with the interviewer |