Interview Preparation

Hiring loops reward predictable judgment: you clarify before you solve, you trade off explicitly, and you connect code to production (tests, rollout, failure). This page is a long-form checklist you can rehearse, not generic motivation.

Master 90% syllabus (SDE3+, your stack)

EliteCoach-style prep map: DSA, system design (HLD/LLD), behavioral—linked to this site. Includes drill format.

Interview prep hub (STAR, 12-week plan, company playbooks)

Behavioral outlines, weekly rhythm, and Big Tech–style playbooks.

DSA + system design mind-map

Pattern library, ~226 problem write-ups, cheatsheets, system-design curriculum.

Generative AI for engineers

LLM systems depth; pair with /dsa/interview-prep/generative-ai for whiteboard pacing.

---

How to use this page

1. Read Core basics once and turn bullets into spoken habits (“I’d clarify X, then sketch two approaches…”).
2. Use Recognition cues before mocks: JD text → themes → proof points you actually have.
3. Run Study cadence blocks on a timer—upgrade quality of talking, not volume of solving.
4. Memorize Scripts loosely; sound like yourself, keep the skeleton.

---

Topic study plan (deep pages)

Each file under /interview/topics/... uses Core details → Understanding → Senior understanding → Diagram.

Author-only scaffold: `content/core-docs/interview/topics/topic-page-template.mdx` (`publishDocs: false`).

---

Core basics — meta signals graders listen for

Senior bar often adds economics: latency tails, infra cost implications, stakeholder trade (“we ship degraded read path first”).

---

Core basics — universal opening moves (≈first 3–5 minutes)

Step 1 — Restate impact

“What the user/org needs” in one plain sentence—not parroting interviewer—then pause for corrections.

Step 2 — Clarify constraints checklist

Verbally scan as appropriate:

Step 3 — Tiny example + invariants

Handwrite a minimal dataset (5–12 elements typical) BEFORE coding/design—catches misunderstanding early.

Define what must stay true each step (“non-decreasing pointers”, “budget never negative”, etc.).

Step 4 — Announce plan

“You’ll hear brute O(n²) baseline, optimized O(n log n) if sort allowed—or O(n) if hashing fits—and I’ll say why assumptions hold.”

Replace with design analog: naive vs scaled decomposition.

---

Understanding — rounds are different games

Live coding interview

Interviewers silently score:

• Pattern/time trade named early.
• Complexity spoken alongside code.
• Edges enumerated while typing (negatives, empty, duplicates, overflow).
• Tests articulated even if partial write.

Anti-pattern: mute coding until “done,” then scramble for edge cases under pressure.

System design interview

Expected arc:

Avoid technology-first ladders (“Kafka + Cassandra”) without query pressure evidence.

Behavioral / leadership

Staff expectations:

Use STAR+: add tradeoff hindsight (“If repeated, shorten validation loop by …”).

GenAI systems interview

Treat as distributed product + stochastic core:

• ingestion & ACLs → retrieval hygiene → guarded tools → eval harness → dashboards → economic routing.

See /gen-ai + /dsa/interview-prep/generative-ai.

---

Recognition cues — read the JD as a blueprint

Translate keywords into prep lanes honestly—don’t retrofit fake stories.

Interviewer style quick mapping

---

Memory hooks

---

Study cadence — sustainable (while employed)

Weekly minimum

Intensive 2 weeks pre-onsite

Daily:

1. 45 min timed solve (new or variant).
2. 20 min cold redo of a miss.
3. 10 min cue-only flashcards (pattern names, numbers only).

Monthly

One full mock crossing coding → design → behavioral; note awkward transitions (where you forget to summarize).

---

Diagrams

Prep feedback loop

Round handoff (keep context)

---

Pitfalls (don’t lose the bar on “soft” failure)

• Silent coding with no testing or complexity narration.
• Design without concrete access pattern numbers (read/write ratio, fan-out).
• Behavioral “we” soup—interviewer cannot see your agency.
• Arguing with hints instead of integrating them as new constraints.
• Ignoring remote collaboration or timezone realities if role is distributed.

---

Interview scripts (adapt to your voice)

Start coding:

“I’ll restate expected behavior, list 2–3 edge cases, then implement the straightforward approach at O(…), and improve if a constraint allows.”

Mid-design:

“Before sharding, I want read/write QPS, payload growth, and whether strong consistency is required for money reads—those decide partition key vs cache strategy.”

Behavioral close:

“If I repeated the project, I’d shorten the validation loop by shipping a dark read on the new path one week earlier—same risk budget, faster signal.”

---

GenAI + interview prep crosswalk