Multimodal & Document AI · 3 to 4 hours
Design a verifiable document and media pipeline
Design a pipeline that turns a partly scanned course pack and a lesson recording into cited study material, tracking coverage, transforms, and cross-modal conflicts, and decide what it may answer while processing is still partial.
4 h
Used in interview rounds:
AI System Design (Hard)1
Step 1: The question
Using the scenario, design the service that turns the course pack and the recording into cited study notes and draft exercises. Explain job states and coverage, evidence links and transforms, the audio and slide disagreement, escalation and review, access checks, evaluation, and release criteria. Then give the system's response to each of the six waiting questions.
Constraints
- Every answer cites a page region or recording time range the asker can open
- Escalation is capped at 2 attempts per item and 30 escalated items per source
- No rule makes the audio or the slide authoritative
- A draft exercise publishes only after reviewer approval, with every key constraint present and settled
Deliverables
- Job identity, states, and coverage accounting for both sources at minute 40
- Evidence links from claims to page regions and time ranges, with the transforms recorded and how a disagreement is stored
- Escalation, review, and release rules, and the evaluation that justifies them
- Parsing isolation, handling of instructions inside sources, and the access check on each derivative
- Traces and failure-injection tests, and how an extraction change is rolled out
- A response to each of the six waiting questions
Hard questions are part of Premium: the study guide, reference answer, rubric, and exercises unlock with Premium.