Build a preparation plan
Map the likely rounds for your role, seniority, and target company.
Open ExplorePrepare for every interview round.
Learn what to do, what evidence to produce, how it is evaluated, and which mistake to remove before the real interview.
Choose where to begin
Start with a plan, a specific interview round, or a task you can practice now.
Study a specific round
Jump to the sequence, evidence, evaluation, and mistakes for that interview.
Choose a roundStart with a task
Browse the active bank by stage, role, seniority, format, and topic.
Browse tasksRound-by-round preparation
Timings vary by company. The work sequence and evaluated evidence are more stable than the exact agenda.
HR and technical screening
Watch or listen to the prompt, then answer aloud under a time limit with live transcription.
20-45 minVoice
Before you start
- Write the target role and the signal this question is likely testing.
- Choose one recent example with your decision, action, and result.
Run the round
- 1Answer the question directly in the first sentence.
- 2Support the answer with one concrete example or technical scenario.
- 3Handle a focused follow-up without contradicting the original answer.
- 4Close by connecting the evidence back to the target role.
Evidence produced
A recording, transcript, answer structure, and attempt history.
How it is evaluated
Clarity, relevance, correctness, ownership, concrete evidence, and seniority-appropriate depth.
Algorithms and platform coding
Implement a solution in a code editor, run examples and checks, inspect failures, and submit the final code.
45-90 minCode
Before you start
- Confirm the language, runtime, allowed libraries, and expected input and output.
- Reserve time for tests, a final walkthrough, and complexity discussion.
Run the round
- 1Restate the contract and identify ambiguous inputs.
- 2Choose an approach, implement it, and keep the code runnable.
- 3Use test failures as evidence, then explain complexity and trade-offs.
- 4Review names, edge cases, and the final output before submitting.
Evidence produced
Source code, test results, runtime output, complexity reasoning, and the submitted revision.
How it is evaluated
Correctness, edge cases, complexity, code quality, debugging process, and adherence to the task contract.
SQL interviews
Inspect the schema and sample data, write a query with schema autocomplete, run it, and review the result table or database error.
30-60 minSQL
Before you start
- State the output grain before writing a join.
- Mark nullable fields, relationship cardinality, and the expected result columns.
Run the round
- 1Translate the requested output into grain, columns, and filters.
- 2Inspect keys and relationships before choosing joins.
- 3Run the query, validate edge cases, and explain its plan-sensitive parts.
- 4Compare the result shape to the request and remove accidental duplicates.
Evidence produced
The executable SQL query, returned rows, database diagnostics, and query feedback.
How it is evaluated
Correctness, output shape, null and duplicate handling, joins and aggregation, and query-plan efficiency.
System and low-level design
Clarify the problem, define requirements, model components or contracts, draw the architecture, and explain trade-offs.
45-60 minDiagram
Before you start
- Prepare a short checklist for users, scale, latency, consistency, privacy, and availability.
- Decide which flow and trade-off you will trace in depth if time becomes limited.
Run the round
- 1Clarify users, scale, constraints, and what is out of scope.
- 2Define requirements and estimates before drawing components.
- 3Trace critical flows, failure paths, and one deep technical trade-off.
- 4Review the design against the original requirements and state remaining risks.
Evidence produced
Clarifying questions, written decisions, diagrams, contracts, failure paths, and the final walkthrough.
How it is evaluated
Scope, capacity reasoning, ownership boundaries, data flow, reliability, trade-offs, and implementation depth.
Behavioral and leadership
Build evidence-backed notes, answer aloud, handle follow-up probes, and compare attempts.
30-60 minVoice
Before you start
- Choose a story with a real decision, tension, and consequence.
- List the facts that prove your own contribution and the result.
Run the round
- 1Choose a real story that contains a decision, tension, and outcome.
- 2Use short notes while answering aloud; do not memorize a script.
- 3Answer follow-ups with evidence and close with what changed afterward.
- 4Review the transcript for team-only language and unsupported claims.
Evidence produced
Story notes, recording, transcript, follow-up answers, and signal-level feedback.
How it is evaluated
Situation clarity, individual ownership, decisions, measurable impact, reflection, and leadership judgment.
Code review, debugging, and incidents
Inspect code, diffs, tests, logs, metrics, or infrastructure evidence; identify the cause; then propose and verify the smallest safe change.
45-75 minInvestigation
Before you start
- Write the observed failure separately from your first hypothesis.
- Define what evidence would confirm or reject that hypothesis.
Run the round
- 1Establish the observed failure and separate facts from hypotheses.
- 2Narrow the cause using code, logs, tests, or metrics.
- 3Apply the smallest safe change and prove both recovery and non-regression.
- 4State the remaining risk, monitoring signal, and rollback trigger.
Evidence produced
Inline review comments, hypotheses, commands, patches, verification output, risks, and rollback steps.
How it is evaluated
Evidence quality, root-cause reasoning, blast-radius control, fix safety, regression protection, and operational judgment.
Machine learning and infrastructure
Work in a domain-specific notebook, experiment console, model incident, CI/CD, Kubernetes, observability, or infrastructure review workspace.
45-90 minSpecialist lab
Before you start
- Identify the baseline, target metric, operating constraints, and failure cost.
- Choose the domain tool that can produce the evidence needed for a decision.
Run the round
- 1Read the complete operational or experimental context before deciding.
- 2Investigate through the domain tool: notebook, dashboard, terminal, pipeline, or plan.
- 3Commit to a recommendation with validation, risk, and rollback criteria.
- 4State what you would monitor after the decision and when you would reverse it.
Evidence produced
Queries, code, metrics, model or system decisions, incident actions, and a final recommendation.
How it is evaluated
Domain correctness, experimental or operational rigor, production constraints, risk management, and measurable validation.
Readiness uses explicit criteria
Each task defines its criteria and seniority expectations. Junior, Middle, and Senior answers can use the same prompt while depth, ownership, trade-offs, and pass thresholds change.
The next task uses attempt evidence
Role, specialization, company environment, topic coverage, previous attempts, and stage readiness shape the next recommendation. Explore and Practice use the same active task bank.
Put the sequence into practice
Browse tasks by interview stage and open the workspace that matches the evidence you need to train.