# Technical Hiring — Structured Interviews That Find Signal

<!-- hint:slides topic="Structured interview pipeline: scorecard, question types, rubrics, independent scoring, and bias mitigation" slides="6" -->

## Why Structured Interviews Matter

Unstructured interviews are **poor predictors** of job performance. They favor charisma, similarity bias, and the "gut feel" that often reflects our own biases. **Structured interviews**—same questions, same rubrics, independent scoring—improve reliability, fairness, and legal defensibility. The goal: **find signal, reduce noise.**

Sources: *Who* (Smart & Street), *Work Rules!* (Laszlo Bock, ex-Google). Both emphasize: define what you need, ask the same questions, score against a rubric.

## Designing a Scorecard

Before writing questions, define the **scorecard**: what competencies matter, and how you'll evaluate them.

**Must-have vs. nice-to-have:**
- **Must-have** — Non-negotiable. Missing one = no hire.
- **Nice-to-have** — Strengths that differentiate; can compensate elsewhere.

**Example scorecard for a mid-level engineer:**

| Competency | Must/Need | How We Assess |
|------------|-----------|---------------|
| Problem-solving | Must | Technical screen, system design |
| Communication | Must | Behavioral, technical (explain reasoning) |
| Collaboration | Must | Behavioral (STAR) |
| Domain experience | Nice | Resume, technical depth |
| Culture add | Nice | Behavioral, values alignment |

**Rule:** Scorecard first. Questions and rubrics flow from it.

## Interview Types

| Type | Purpose | Format |
|------|---------|--------|
| **Behavioral** | Past behavior predicts future | STAR format; same questions for all |
| **Technical** | Technical skill, problem-solving | Coding, debugging, or architecture; clear rubric |
| **System design** | Architecture, trade-offs, scale | Whiteboard or diagram; scenario-based |
| **Take-home** | Async signal; work sample | Bounded task; blind or semi-blind review |

**Balance the mix:** Don't over-index on any one. Behavioral + technical + one other is typical.

## Writing Good Questions

### Behavioral: STAR Format

- **S**ituation — Context
- **T**ask — Their responsibility
- **A**ction — What they did
- **R**esult — Outcome, what they learned

**Example:** "Tell me about a time you disagreed with a technical decision. What was the situation, what did you do, and what was the result?"

**Rubric:** Strong answer = specific situation, clear ownership, concrete action, reflective outcome. Weak = vague, someone else did it, no reflection.

### Technical: Clear Rubric

- **Question:** Bounded, unambiguous. "Given an array of integers, find two that sum to target."
- **Rubric:** What does strong/medium/weak look like? Edge cases? Optimization?
- **Follow-ups:** Pre-prepared so interviewers probe consistently.

**Example rubric:** Strong = correct solution, handles edge cases, discusses complexity. Medium = correct with hint. Weak = stuck, wrong approach, can't explain.

## Conducting the Interview

1. **Rapport (2–3 min)** — Brief intro, set expectations, put them at ease.
2. **Calibration** — If multiple interviewers, align on rubric before the day.
3. **Note-taking** — Take notes *during*; don't rely on memory. Quote key phrases.
4. **Time management** — Reserve 5–10 min for their questions.
5. **No leading** — Don't hint at answers; let them perform.

**Candidate experience:** Start and end on time. Explain what's next. Thank them.

## Evaluation: Independent Scoring and Debrief

1. **Independent scoring** — Each interviewer scores *before* the debrief. No groupthink.
2. **Debrief meeting** — Compare scores, discuss evidence, resolve discrepancies.
3. **Hire/no-hire** — Based on scorecard; must-haves must be met.
4. **Document** — Notes and scores for legal and calibration.

**Calibration:** Periodically review: Are we too lenient? Too harsh? Use data.

## Common Biases

| Bias | What It Is | Mitigation |
|------|------------|------------|
| **Halo effect** | One strength overshadows weaknesses | Score each competency separately |
| **Similarity bias** | Prefer people like us | Structured questions; diverse panel |
| **Anchoring** | First impression dominates | Independent scoring; evidence-based |
| **Recency** | Remember the end of the interview | Take notes throughout |
| **Confirmation** | Seek evidence that matches initial view | Rubric first; look for disconfirming evidence |

**Inclusive hiring:** Reduce unnecessary requirements (e.g., "CS degree"); focus on competency. Diverse interview panels. Bias training.

## Candidate Experience

- **Clear communication** — What to expect, how long, who they'll meet.
- **Respect their time** — Start on time; don't overrun.
- **Feedback** — If you reject, provide brief constructive feedback (where legally appropriate).
- **Speed** — Don't leave them hanging for weeks.

## Hiring Pipeline Flow

```mermaid
flowchart TD
    A[Define scorecard] --> B[Screen resumes]
    B --> C[Phone/technical screen]
    C --> D[On-site: Behavioral]
    C --> E[On-site: Technical]
    C --> F[On-site: System design]
    D --> G[Independent scoring]
    E --> G
    F --> G
    G --> H[Debrief meeting]
    H --> I{Hire/No-hire}
    I -->|Hire| J[Offer]
    I -->|No-hire| K[Reject + feedback]

    style A fill:#3fb950
    style G fill:#58a6ff
    style H fill:#f85149
```

## Key Takeaways

1. **Structured > unstructured** — Same questions, same rubric, independent scoring.
2. **Scorecard first** — Define competencies before writing questions.
3. **STAR for behavioral** — Situation, Task, Action, Result.
4. **Clear rubrics for technical** — What does strong/medium/weak look like?
5. **Mitigate bias** — Separate scoring, diverse panel, evidence-based.
6. **Candidate experience** — Clear, respectful, timely.