Address miscommunication and alignment

# Model answer (STAR) + playbook ## Example STAR story (interviewer misalignment) - Situation: In a technical screen, I was asked to find the k-th largest value from a data stream. I assumed the input could be fully materialized in memory and started with a sort-based solution. - Task: Deliver a correct, efficient solution under unspecified constraints while communicating clearly. - Action (initial approach): I outlined an O(n log n) approach by collecting all values and sorting. I started coding. - Detecting misalignment: The interviewer asked, "How would this handle 10 million elements?" and "What if the stream never ends?" Those questions signaled that memory/streaming constraints mattered, and my assumptions were off. - Clarifying: I paused and explicitly asked: - Is the stream unbounded and ongoing? - Is there a memory limit (e.g., O(k) or O(1) extra space)? - Is the answer required to be exact, or is approximation acceptable? - What are typical ranges for n and k? We aligned on: unbounded stream, exact result, O(k) memory preferred, k ≤ 1,000. - Adjusting approach: I switched to maintaining a size-k min-heap: - Push each number; if heap size > k, pop the smallest. - The root is the k-th largest at any time. - Complexity: O(n log k) time, O(k) space, which meets the constraints. - I walked through a quick example for k=3 with stream [5,1,7,3,9,2]: final heap [5,7,9] ⇒ k-th largest = 5. - Result: We finished within time, and the interviewer noted the pivot and clarity as positives. My takeaway was to confirm constraints up front and checkpoint alignment. ## Why this works - It demonstrates early detection (interviewer probes), explicit clarification (constraints, acceptance criteria), a principled pivot (heap vs sort), and measurable impact (O(n log k) vs O(n log n) and O(k) memory). ## Tactics to validate assumptions and align in future screens 1. 60–90 second problem restatement - Restate the goal, input/output shape, and success criteria in your own words. - Ask: "Any constraints on time/space? Streaming vs batch? Exact vs approximate?" 2. Examples-first alignment - Propose 2–3 test cases, including one edge/extreme (e.g., k=1, duplicates, very large n). Confirm with the interviewer. 3. Explicit assumption log - Say assumptions out loud and label them: "Assumption: input may exceed memory; please confirm." Convert assumptions to facts or adjust. 4. Complexity targets upfront - If not specified, volunteer a target with trade-offs: "I can do O(n log k) time, O(k) space with a min-heap; alternatively O(n log n) with full sort. Which aligns better with constraints?" 5. Periodic checkpoints - After sketching the high-level approach and after pseudocode, ask: "Does this direction match your intent before I implement?" 6. Constraint drill-down (fast checklist) - Input size ranges, memory limits, mutability, ordering guarantees, duplicates, failure modes, concurrency, time budget. 7. Misalignment signals to watch for - Repeated questions about constraints you haven’t addressed, new edge cases that break your plan, or the interviewer asking "What if it doesn’t fit in memory?" Treat these as prompts to pause and realign. 8. Pivot gracefully (3-step script) - Acknowledge: "I see my approach assumes full materialization." - Reframe: "Given an unbounded stream and O(k) memory, a min-heap fits better." - Plan: "I’ll implement the heap, prove correctness on a small example, then discuss complexity." 9. Validate with quick tests + invariants - Run through sample inputs and extremes; state invariants (e.g., "Heap maintains the k largest seen so far"). Verify big-O and space explicitly. 10. Timebox and iterate - Keep clarifying questions to ~1 minute initially; start building; iterate with checkpoints rather than over-clarifying. ## Common pitfalls and how to avoid them - Pitfall: Overfitting to a specific data structure suggestion. Avoid leading questions like "Do you want a heap?" Instead, present options with trade-offs. - Pitfall: Silent assumptions (e.g., finite input). Say them explicitly and confirm. - Pitfall: Spending too long on meta-communication. Timebox, implement, and re-check alignment. ## Mini script you can reuse - Opening: "Let me restate the problem and constraints I’m hearing... Are there memory/time targets? Is streaming required?" - Before coding: "Two approaches: full sort O(n log n), or min-heap O(n log k), O(k) space. Which aligns with your intent?" - During: "Quick checkpoint: does this plan match what you had in mind?" - On signal of misalignment: "Sounds like my assumption about memory was off. I’ll pivot to a heap-based streaming design and walk through an example." These habits demonstrate situational awareness, efficient communication, and the ability to course-correct under pressure—key behaviors interviewers look for in a technical screen.