Senior System Architecture
When to Use
- Define or review cross-service/system boundaries and contracts
- Compare architectural options with explicit trade-offs and NFR impact
- Author or review ADRs for decisions that are hard to reverse
- Run architecture review before major launches or vendor commitments
- Plan migration off legacy systems (strangler, parallel run, cutover)
- Establish architecture principles, standards, and exception process
When NOT to Use
- Single-team service RFC and implementation slices →
senior-software-engineer
- Lakehouse, mesh, or enterprise data modeling →
data-architect
- VPC, Kubernetes, Terraform, or CI/CD build →
infrastructure-engineer, devops
- IDP, golden paths, Backstage →
platform-engineer
- Milestones, RAID, steering status →
technical-program-manager
- Release cutover tactics only →
deployment-strategist
- Security control catalog and enterprise GRC →
cybersecurity
- LLM/RAG system design →
ai-engineer
- Business strategy, issue trees, steerCo cases →
business-consultant
- Applied AI / LLM solution architecture →
applied-ai-architect-commercial-enterprise
- Customer-facing solution design, RFP/RFI, PoC scope, deal integration →
solutions-architect
Related skills
| Need |
Skill |
| Service-level RFC and code review |
senior-software-engineer |
| Data domain and governance architecture |
data-architect |
| Cloud solution and migration architecture |
cloud-architect |
| Enterprise cloud governance and landing zones |
enterprise-cloud-architect |
| Cloud/network/IaC delivery |
infrastructure-engineer, cloud-engineer |
| Platform-as-product and golden paths |
platform-engineer |
| Multi-team launch coordination |
technical-program-manager |
| Rollout and rollback planning |
deployment-strategist |
| Security architecture and controls |
cybersecurity, information-security-engineer |
| Requirements and business constraints |
business-analyst |
| Strategy, business case, operating model |
business-consultant |
| AI/LLM solution patterns |
ai-engineer |
| Commercial/enterprise applied AI architecture |
applied-ai-architect-commercial-enterprise |
| Customer deal solution, RFP, PoC handoff |
solutions-architect |
Core Workflows
1. Frame the decision
Capture before drawing boxes:
- Business outcome and measurable success criteria
- Constraints: budget, timeline, compliance, existing estate
- Non-goals (explicit scope cuts)
- Stakeholders and decision owner
- Reversibility: one-way door vs two-way door
One-way doors require ADR + architecture review. Two-way doors can stay in team RFC.
2. Model the system (C4-lite)
Minimum views for reviewers:
- Context — actors, external systems, trust zones
- Containers — deployable units, data stores, queues, who owns each
- Critical path — sequence diagram for highest-risk flows only
Label every arrow: sync/async, protocol, auth model, and failure behavior.
See references/integration_patterns.md.
3. Define NFRs and quality attributes
For each capability, specify targets (not vague "high availability"):
| Attribute |
Example target |
Verification |
| Availability |
99.9% monthly |
SLO, error budget |
| Latency |
p99 < 300ms read |
Load test + prod SLO |
| Throughput |
5k RPS peak |
Capacity model |
| Durability |
RPO 1h, RTO 4h |
DR drill |
| Security |
mTLS east-west, OIDC |
Threat model link |
| Cost |
<$X / 1M requests |
FinOps estimate |
See references/nfr_quality_attributes.md.
4. Evaluate options
Present at least two viable options plus "do nothing / minimal change":
| Criterion |
Weight |
Option A |
Option B |
| Time to value |
|
|
|
| Operational burden |
|
|
|
| Scalability headroom |
|
|
|
| Team skill fit |
|
|
|
| Vendor lock-in |
|
|
|
| Security/compliance fit |
|
|
|
Recommend one; document rejected options and why.
See references/adr_template.md.
5. Architecture review
Before build or contract signature:
- Problem and constraints restated in one paragraph
- Diagrams current; contracts versioned (OpenAPI, event schema)
- Failure modes: partial outage, dependency down, poison messages
- Data: ownership, retention, PII, migration path
- Observability: golden signals per container
- Security: authz boundaries, secrets, blast radius
- Rollout and rollback linked to
deployment-strategist if multi-phase
See references/architecture_review.md.
6. Evolution and migration
For legacy replacement:
- Identify capability slices that can move independently
- Prefer strangler over big-bang when risk is high
- Define parity criteria before cutover
- Plan dual-write / dual-read duration and reconciliation
- Deprecate old path with telemetry proving zero traffic
See references/migration_evolution.md.
When to load references
- ADR format and bar →
references/adr_template.md
- Review checklist →
references/architecture_review.md
- Sync, events, sagas, APIs →
references/integration_patterns.md
- SLOs, capacity, DR, cost →
references/nfr_quality_attributes.md
- Strangler and cutover →
references/migration_evolution.md
1---2name: senior-system-architecture3description: Guides senior system and solution architecture—cross-service boundaries, integration patterns, non-functional requirements (scale, reliability, security, cost), ADRs, C4-style modeling, architecture review, build-vs-buy, and phased migration (strangler, dual-write). Use when designing multi-service systems, evaluating platform or vendor choices, writing or reviewing architecture decision records, defining standards and principles, or assessing technical risk across domains—not for single-service RFCs and module design (senior-software-engineer), data platform or mesh decisions (data-architect), cloud landing zone, Well-Architected, and migration architecture (cloud-architect), cloud/IaC implementation (infrastructure-engineer, cloud-engineer), internal developer platform product (platform-engineer), or program tracking (technical-program-manager). For business strategy and cases, use business-consultant; for applied AI (RAG, agents, copilots), use applied-ai-architect-commercial-enterprise.4---56# Senior System Architecture78## When to Use910- Define or review cross-service/system boundaries and contracts11- Compare architectural options with explicit trade-offs and NFR impact12- Author or review ADRs for decisions that are hard to reverse13- Run architecture review before major launches or vendor commitments14- Plan migration off legacy systems (strangler, parallel run, cutover)15- Establish architecture principles, standards, and exception process1617## When NOT to Use1819- Single-team service RFC and implementation slices → `senior-software-engineer`20- Lakehouse, mesh, or enterprise data modeling → `data-architect`21- VPC, Kubernetes, Terraform, or CI/CD build → `infrastructure-engineer`, `devops`22- IDP, golden paths, Backstage → `platform-engineer`23- Milestones, RAID, steering status → `technical-program-manager`24- Release cutover tactics only → `deployment-strategist`25- Security control catalog and enterprise GRC → `cybersecurity`26- LLM/RAG system design → `ai-engineer`27- Business strategy, issue trees, steerCo cases → `business-consultant`28- Applied AI / LLM solution architecture → `applied-ai-architect-commercial-enterprise`29- Customer-facing solution design, RFP/RFI, PoC scope, deal integration → `solutions-architect`3031## Related skills3233| Need | Skill |34|---|---|35| Service-level RFC and code review | `senior-software-engineer` |36| Data domain and governance architecture | `data-architect` |37| Cloud solution and migration architecture | `cloud-architect` |38| Enterprise cloud governance and landing zones | `enterprise-cloud-architect` |39| Cloud/network/IaC delivery | `infrastructure-engineer`, `cloud-engineer` |40| Platform-as-product and golden paths | `platform-engineer` |41| Multi-team launch coordination | `technical-program-manager` |42| Rollout and rollback planning | `deployment-strategist` |43| Security architecture and controls | `cybersecurity`, `information-security-engineer` |44| Requirements and business constraints | `business-analyst` |45| Strategy, business case, operating model | `business-consultant` |46| AI/LLM solution patterns | `ai-engineer` |47| Commercial/enterprise applied AI architecture | `applied-ai-architect-commercial-enterprise` |48| Customer deal solution, RFP, PoC handoff | `solutions-architect` |4950## Core Workflows5152### 1. Frame the decision5354Capture before drawing boxes:5556- Business outcome and measurable success criteria57- Constraints: budget, timeline, compliance, existing estate58- Non-goals (explicit scope cuts)59- Stakeholders and decision owner60- Reversibility: one-way door vs two-way door6162**One-way doors** require ADR + architecture review. **Two-way doors** can stay in team RFC.6364### 2. Model the system (C4-lite)6566Minimum views for reviewers:67681. **Context** — actors, external systems, trust zones692. **Containers** — deployable units, data stores, queues, who owns each703. **Critical path** — sequence diagram for highest-risk flows only7172Label every arrow: sync/async, protocol, auth model, and failure behavior.7374**See `references/integration_patterns.md`.**7576### 3. Define NFRs and quality attributes7778For each capability, specify targets (not vague "high availability"):7980| Attribute | Example target | Verification |81|---|---|---|82| Availability | 99.9% monthly | SLO, error budget |83| Latency | p99 < 300ms read | Load test + prod SLO |84| Throughput | 5k RPS peak | Capacity model |85| Durability | RPO 1h, RTO 4h | DR drill |86| Security | mTLS east-west, OIDC | Threat model link |87| Cost | <$X / 1M requests | FinOps estimate |8889**See `references/nfr_quality_attributes.md`.**9091### 4. Evaluate options9293Present at least two viable options plus "do nothing / minimal change":9495| Criterion | Weight | Option A | Option B |96|---|---|---|---|97| Time to value | | | |98| Operational burden | | | |99| Scalability headroom | | | |100| Team skill fit | | | |101| Vendor lock-in | | | |102| Security/compliance fit | | | |103104Recommend one; document rejected options and why.105106**See `references/adr_template.md`.**107108### 5. Architecture review109110Before build or contract signature:1111121. Problem and constraints restated in one paragraph1132. Diagrams current; contracts versioned (OpenAPI, event schema)1143. Failure modes: partial outage, dependency down, poison messages1154. Data: ownership, retention, PII, migration path1165. Observability: golden signals per container1176. Security: authz boundaries, secrets, blast radius1187. Rollout and rollback linked to `deployment-strategist` if multi-phase119120**See `references/architecture_review.md`.**121122### 6. Evolution and migration123124For legacy replacement:1251261. Identify **capability slices** that can move independently1272. Prefer **strangler** over big-bang when risk is high1283. Define **parity criteria** before cutover1294. Plan **dual-write / dual-read** duration and reconciliation1305. Deprecate old path with telemetry proving zero traffic131132**See `references/migration_evolution.md`.**133134## When to load references135136- **ADR format and bar** → `references/adr_template.md`137- **Review checklist** → `references/architecture_review.md`138- **Sync, events, sagas, APIs** → `references/integration_patterns.md`139- **SLOs, capacity, DR, cost** → `references/nfr_quality_attributes.md`140- **Strangler and cutover** → `references/migration_evolution.md`
daemon-blockint-tech