Mock Interview

Step 2: Run each round

For each round, ask the question and wait for the answer. Apply the same Socratic evaluation as the interview-qa skill — probe once before scoring.

Round 1 — Warm-up question bank (pick one):

• "Walk me through your experience with ML in production."
• "What's the most challenging data problem you've worked on?"
• "How do you decide which model to use for a new problem?"

Round 2 — Statistics (pick 2 from different subtopics):

• p-value interpretation, Type I/II errors
• Central Limit Theorem and when it applies
• Bayesian vs frequentist — practical difference
• Designing an A/B test from scratch
• Explaining a confidence interval to a non-technical stakeholder

Round 3 — ML Theory (pick 2):

• Explain the bias-variance tradeoff with a concrete example
• How does XGBoost differ from a random forest?
• What is regularization and why does it work?
• How would you handle a severely imbalanced dataset?
• Explain how backpropagation works intuitively

Round 4 — Technical (pick one based on user's weaker area from progress log):

SQL option:

"Given a table events(user_id, event_type, timestamp), write a query to find all users who performed event type 'purchase' within 7 days of performing event type 'signup'."

Coding option:

"Implement logistic regression from scratch using only numpy. Include fit() and predict() methods."

Round 5 — System Design (pick one):

• "Design a real-time fraud detection system for an e-commerce platform."
• "How would you build a recommendation engine for a streaming service?"
• "Design the ML pipeline for a churn prediction model at a SaaS company."

Step 3: Mid-interview behavior

• Say "Thank you" briefly after each answer before probing — this mimics real interview pacing
• Keep track of time context: after Round 2, say "We're about 8 minutes in."
• If the user is really struggling, give one hint phrased as "Some candidates approach this by..."
• Don't break character except to ask if they want to pause

Step 4: Generate final scorecard

After all rounds:

═══════════════════════════════════════
       MOCK INTERVIEW SCORECARD
═══════════════════════════════════════

Candidate: [User]
Date: <date>
Role: Data Scientist / ML Engineer

SCORES
──────────────────────────────────────
Round 1  Warm-up          <score>/5
Round 2a Statistics       <score>/5
Round 2b Statistics       <score>/5
Round 3a ML Theory        <score>/5
Round 3b ML Theory        <score>/5
Round 4  Technical        <score>/5
Round 5  System Design    <score>/5
──────────────────────────────────────
OVERALL                   <avg>/5

HIRING SIGNAL
──────────────────────────────────────
4.5–5.0  → Strong hire
3.5–4.4  → Hire (with some reservations)
2.5–3.4  → Mixed — more prep needed
Below 2.5 → Not ready yet

Your signal: <label>

TOP STRENGTH
<one sentence on what they did best>

CRITICAL GAP
<the single most important thing to improve>

PRIORITY TOPICS FOR NEXT SESSION
1. <topic> — reason
2. <topic> — reason
═══════════════════════════════════════

Step 5: Log to progress

Append a mock interview summary entry to memory/progress.json:

{
  "timestamp": "<ISO>",
  "topic": "mock-interview",
  "question": "Full mock interview — 7 questions",
  "score": <overall avg>,
  "note": "<hiring signal> — strength: <X>, gap: <Y>"
}

Then commit:

git add memory/progress.json
git commit -m "mock-interview: scored <avg>/5 — <hiring signal> — <date>"