Original practice scenario: A hypothetical semiconductor-equipment analytics interface changes a pressure field from one unit to another and adds a chamber identifier. Design a compatible event contract.
Read model answer +
Version semantic changes explicitly and validate event context before calculation. Keeping the same numeric field name after a unit change can corrupt analysis without triggering a parser failure.
Work through it
Define required equipment, chamber, event-time and measurement fields with explicit units and schema version, distinguishing unknown chamber mapping from a valid shared-tool event.
Implement a reviewed adapter for supported old versions, retaining original values and conversion provenance rather than silently accepting every shape.
Reject ambiguous versions, preserve idempotent event identities and test unit boundaries, missing chamber references and replayed events under the new consumer.
Tradeoffs to discuss
Adapters reduce producer coordination but increase supported-state complexity; a clean cutover simplifies consumers while requiring stronger rollout control.
Common pitfalls
Do not infer units from a typical numeric range.
A newly required chamber dimension may reveal that historical tool-level records cannot be honestly assigned to one chamber.
Original practice scenario: A hypothetical wafer-processing tool has several chambers, setup times and product restrictions. A team claims a scheduling change improves throughput. Design an evaluation before assuming the simulation is trustworthy.
Read model answer +
Validate the simulation's representation before comparing policies. Throughput gains can arise from omitted setup, downtime or eligibility constraints rather than a genuinely better schedule.
Work through it
Define lots, processing stages, chamber capabilities and setup transitions, checking conservation of work-in-progress and resource occupancy on small hand-worked cases.
Calibrate timing inputs from appropriate evidence and replay a held-out workload, comparing queue lengths, completion times and bottlenecks with observed or independently specified behavior.
Run policies under identical demand and random seeds where appropriate, varying uncertain inputs and reporting tail delays and starvation alongside throughput.
Tradeoffs to discuss
Detailed simulation improves realism but requires more uncertain parameters; a simpler model is easier to audit while omitting interactions.
Common pitfalls
Do not let a policy select a chamber that the product cannot use.
A simulation comparison is conditional evidence and does not authorize real equipment control.
