recruiting-ops

Files (4)

• README.md
• SKILL.md
• evals.json
• job-description-templates.md · references/

SKILL.md (first lines)

## Key principles

1. **Structured process reduces bias** - Unstructured interviews measure confidence and
   likability, not job performance. Every hiring decision should rest on a defined
   scorecard with consistent signals evaluated the same way across all candidates.
   Standardize questions, calibrate interviewers, and separate data collection from
   evaluation to prevent halo effects and affinity bias.

2. **Speed is a competitive advantage** - The best candidates are off the market in
   7-14 days. Slow loops, delayed feedback, and scheduling gaps lose top talent to
   faster competitors. Measure time-in-stage, eliminate unnecessary interview rounds,
   and make offer decisions within 24 hours of a final interview.

3. **Sell while you evaluate** - Every touchpoint is a chance to lose or win a candidate.
   Interviewers who show up unprepared, ask hostile questions, or fail to explain the
   role's impact drive rejection rates up. Train every interviewer on the pitch: mission,
   team, growth path, and why this role matters now.

4. **Data-driven pipeline** - Track conversion rates at every funnel stage. If
   application-to-screen rate is high but screen-to-onsite is low, the phone screen is
   miscalibrated. If offer acceptance rate is below 80%, the offer or closing process
   is broken. Metrics tell you where the process leaks before it costs you headcount.

5. **Candidate experience matters** - Candidates who have a bad experience - ghosting,
   rude interviewers, confusing processes - become vocal detractors. Candidates who have
   a great experience become fans, even if not hired. Timely communication, clear
   expectations, and respectful feedback are the baseline.

---

## Core concepts

### Recruiting funnel

```
Sourced / Applied
      |
  Screened (resume + application review)
      |
  Phone / Video Screen
      |
  Technical / Skills Screen  (optional, role-dependent)
      |
  Onsite / Final Loop
      |
  Reference Check
      |
  Offer Extended
      |
  Offer Accepted
      |
  Day 1
```

Each stage has a target conversion rate. Deviation from baseline signals a broken
stage, not a broken candidate pool. Track volume and conversion at every gate.

### Sourcing channels

| Channel | Best for | Cost | Time to fill |
|---|---|---|---|
| Employee referrals | Culture fit, passive candidates | Low | Fast |
| LinkedIn Recruiter | Senior / specialized roles | Medium | Medium |
| Job boards (LinkedIn, Indeed, Greenhouse) | High-volume, entry/mid-level | Low-Medium | Fast |
| Niche communities (Discord, Slack, forums) | Technical / domain-specific roles | Low | Slow |
| Recruiting agencies | Executive, urgent, highly specialized | High | Variable |
| GitHub / Dribbble / portfolio sites | Engineers, designers | Low | Slow |
| Conferences and meetups | Senior, passive, community-active talent | Medium | Slow |

**Sourcing rule:** Use at least three channels per role. Referrals should be one channel
but never the only channel - they homogenize the candidate pool.

### Screening criteria

A screening rubric must be defined before outreach begins. It contains:

- **Must-haves:** Hard requirements. Failing any disqualifies. Keep this list short (3-5 items).
- **Strong-to-haves:** Differentiating signals that raise conviction. Not disqualifying.
- **Anti-signals:** Patterns that suggest misalignment with the role or team.
- **Selling points:** What to emphasize to this candidate profile to drive conversion.

**Calibration rule:** Every must-have must be directly tied to a core job responsibility.
"5+ years of experience" is a proxy, not a criterion. Replace proxies with skills
or demonstrated behaviors wherever possible.

### Interview loops

An interview loop maps each competency to an interviewer and a set of defined questions.
No interviewer should evaluate a competency they were not assigned. No competency
should be left unassigned.

```
Loop design steps:
1. List 5-7 competencies required for success in the role.
2. Assign each competency to exactly one interviewer.
3. Write 3-5 behavioral or technical questions per competency.
4. Define what a strong, acceptable, and weak response looks like.
5. Hold a pre-loop calibration with all interviewers before the first candidate.
6. Hold a debrief within 24 hours of each loop. Collect written scores first,
   then discuss to prevent anchoring.
```

---

## Common tasks

### Write an inclusive job description

**Template structure:**

```
[ROLE TITLE]

About [Company]:
  2-3 sentences. Mission, stage, and what makes this team worth joining.
  Avoid superlatives ("best", "world-class"). Use specific, factual claims.

What you will do:
  5-7 bullet points. Focus on impact, not activities.

From mkurman/zorai, MIT.

More from mkurman/zorai