Delegation & Spreading Context

Tech leads who hoard decisions become bottlenecks and bus-factor risks. Delegation assigns ownership with enough context - ADRs, runbooks, and PR intent - that others can ship and operate Python services without the lead in every thread.

Recipe

Quick-reference recipe card - copy-paste ready.

## Delegation brief
**Outcome:** what success looks like
**Constraints:** SLO, deadline, non-goals
**Context links:** RFC, ADR, dashboard
**Decision authority:** what delegate can choose alone
**Check-in:** Tuesday async status in ticket

When to reach for this:

Lead is reviewer on every PR
Only one engineer knows billing module
Vacation blocks releases
Team growth requires new owners

Working Example

## Delegation: ENG-550 Celery retry policy
 
**Outcome:** Billing tasks retry transient DB errors with jitter; DLQ after 5 attempts.
 
**Constraints:** No API contract change; ship behind flag `BILLING_RETRY_V2`.
 
**Context:**
- ADR-008 queue choice
- Incident INC-9011 post-mortem
- `src/workers/billing.py` current behavior
 
**Authority:** Delegate chooses backoff constants after spike; escalate if new broker needed.
 
**Check-in:** Demo in guild Friday week 2.

# Owner documents context at module top after delegation complete
"""billing_tasks.py
 
Owner: @jamie (delegated 2026-07-01)
Retry policy per ENG-550; DLQ queue billing_dlq.
On-call: pause workers if depth > 50k (runbook orders-api.md).
"""

What this demonstrates:

Outcome and constraints clearer than "fix retries"
Linked artifacts spread context without meetings
Explicit decision authority prevents escalation loops
Module ownership visible to on-call

Deep Dive

How It Works

Outcome delegation - "Reduce CFR for billing" beats "change tenacity config."
Written context - ADRs, runbooks, and README beat tribal knowledge.
Rotation - Release captain, on-call, and CODEOWNERS rotate quarterly.
Graduated authority - Juniors decide implementation; leads own architecture veto.
Feedback loop - Delegate owns retrospective learnings, not just delivery.

Bus-Factor Mitigation

Risk	Mitigation
One deployer	Rotate release captain
One module expert	Pair on next feature; second CODEOWNER
One on-call brain	Runbooks + shadow shifts
One RFC author	Co-author requirement for large changes

Python Notes

# CODEOWNERS excerpt
/src/billing/     @jamie @alex
/docs/runbooks/  @oncall-rotation

Gotchas

Delegating without authority - Assignee must ask lead on every choice. Fix: Document what they can decide solo.
Drive-by delegation in Slack - No ticket, no context. Fix: Delegation brief in tracker.
Taking back work mid-flight - Undermines growth. Fix: Coach via questions unless SEV1.
No ownership transfer - Delegate ships but lead still paged. Fix: Update CODEOWNERS and runbook name.
Context in lead's head only - Vacation outage. Fix: Require PR/RFC links before work starts.

Alternatives

Alternative	Use When	Don't Use When
Shared ownership pair	Critical domain transition	Long-term two owners without split
Guild documentation sprint	Context debt burst	Urgent feature deadline
Staff engineer embed	Cross-squad standards	Simple module handoff
Contractor with handoff doc	Temporary capacity	No internal owner assigned

FAQs

How much context is enough?

Enough that delegate could explain trade-offs to another engineer without you present.

Delegate to juniors?

Yes with smaller scope, buddy reviewer, and clear escalation path.

How avoid abdication?

Delegation includes check-ins and quality bar; abdication is assign and disappear.

What if delegate fails?

Post-mortem on support gaps; adjust brief and pairing - not punishment if learning gap.

Spread context in monorepo?

Package README, ADR links, and CODEOWNERS per top-level module.

Remote delegation?

Written brief essential; record short loom only for complex system tours.

How long rotate on-call?

Weekly primary; document handoff note every Monday.

Delegate incident commander?

Train IC rotation among mids; lead coaches first incidents.

Context vs documentation debt?

Delegate doc updates as part of outcome when touching module.

When lead stays involved?

Architecture veto, security sign-off, and stakeholder comms - not every line of code.

Mentoring & Leveling - grow delegates
Architecture Decision Records (ADRs) - persistent context
On-Call Health - rotate operations
Release Management - release captain rotation
Contribution Guidelines - CODEOWNERS norms

Stack versions: This page was written for Python 3.14.0 (stable 3.14, maintenance 3.13), FastAPI 0.115+, Django 5.2, Flask 3.1, Pydantic 2, PyTorch 2.6+, pandas 2.2+, Polars 1.x, ruff 0.9+, and uv 0.6+.

Recipe

Quick-reference recipe card - copy-paste ready.

## Delegation brief
**Outcome:** what success looks like
**Constraints:** SLO, deadline, non-goals
**Context links:** RFC, ADR, dashboard
**Decision authority:** what delegate can choose alone
**Check-in:** Tuesday async status in ticket

When to reach for this:

Lead is reviewer on every PR
Only one engineer knows billing module
Vacation blocks releases
Team growth requires new owners

Working Example

## Delegation: ENG-550 Celery retry policy
 
**Outcome:** Billing tasks retry transient DB errors with jitter; DLQ after 5 attempts.
 
**Constraints:** No API contract change; ship behind flag `BILLING_RETRY_V2`.
 
**Context:**
- ADR-008 queue choice
- Incident INC-9011 post-mortem
- `src/workers/billing.py` current behavior
 
**Authority:** Delegate chooses backoff constants after spike; escalate if new broker needed.
 
**Check-in:** Demo in guild Friday week 2.

# Owner documents context at module top after delegation complete
"""billing_tasks.py
 
Owner: @jamie (delegated 2026-07-01)
Retry policy per ENG-550; DLQ queue billing_dlq.
On-call: pause workers if depth > 50k (runbook orders-api.md).
"""

What this demonstrates:

Outcome and constraints clearer than "fix retries"
Linked artifacts spread context without meetings
Explicit decision authority prevents escalation loops
Module ownership visible to on-call

Deep Dive

How It Works

Outcome delegation - "Reduce CFR for billing" beats "change tenacity config."
Written context - ADRs, runbooks, and README beat tribal knowledge.
Rotation - Release captain, on-call, and CODEOWNERS rotate quarterly.
Graduated authority - Juniors decide implementation; leads own architecture veto.
Feedback loop - Delegate owns retrospective learnings, not just delivery.

Bus-Factor Mitigation

Risk	Mitigation
One deployer	Rotate release captain
One module expert	Pair on next feature; second CODEOWNER
One on-call brain	Runbooks + shadow shifts
One RFC author	Co-author requirement for large changes

Python Notes

# CODEOWNERS excerpt
/src/billing/     @jamie @alex
/docs/runbooks/  @oncall-rotation

Gotchas

Delegating without authority - Assignee must ask lead on every choice. Fix: Document what they can decide solo.
Drive-by delegation in Slack - No ticket, no context. Fix: Delegation brief in tracker.
Taking back work mid-flight - Undermines growth. Fix: Coach via questions unless SEV1.
No ownership transfer - Delegate ships but lead still paged. Fix: Update CODEOWNERS and runbook name.
Context in lead's head only - Vacation outage. Fix: Require PR/RFC links before work starts.

Alternatives

Alternative	Use When	Don't Use When
Shared ownership pair	Critical domain transition	Long-term two owners without split
Guild documentation sprint	Context debt burst	Urgent feature deadline
Staff engineer embed	Cross-squad standards	Simple module handoff
Contractor with handoff doc	Temporary capacity	No internal owner assigned

FAQs

How much context is enough?

Enough that delegate could explain trade-offs to another engineer without you present.

Delegate to juniors?

Yes with smaller scope, buddy reviewer, and clear escalation path.

How avoid abdication?

Delegation includes check-ins and quality bar; abdication is assign and disappear.

What if delegate fails?

Post-mortem on support gaps; adjust brief and pairing - not punishment if learning gap.

Spread context in monorepo?

Package README, ADR links, and CODEOWNERS per top-level module.

Remote delegation?

Written brief essential; record short loom only for complex system tours.

How long rotate on-call?

Weekly primary; document handoff note every Monday.

Delegate incident commander?

Train IC rotation among mids; lead coaches first incidents.

Context vs documentation debt?

Delegate doc updates as part of outcome when touching module.

When lead stays involved?

Architecture veto, security sign-off, and stakeholder comms - not every line of code.

Mentoring & Leveling - grow delegates
Architecture Decision Records (ADRs) - persistent context
On-Call Health - rotate operations
Release Management - release captain rotation
Contribution Guidelines - CODEOWNERS norms

Stack versions: This page was written for Python 3.14.0 (stable 3.14, maintenance 3.13), FastAPI 0.115+, Django 5.2, Flask 3.1, Pydantic 2, PyTorch 2.6+, pandas 2.2+, Polars 1.x, ruff 0.9+, and uv 0.6+.

Delegation & Spreading Context

Search Documentation

Delegation & Spreading Context