¶ INF-01b — Infrastructure Support & Maintenance
Owner: DevOps Lead — Abdul Moiz
Contributors: AI Engineers (first-line incident reporting) ·
HoD (escalation) · PM (client-side coordination)
Department: D5 — Infrastructure & Compliance
Entity: PrivateAI
Status: Recommended draft · Pending DevOps Lead (Moiz) review
Version: 0.1 | Last updated: 2026-07-18
Classification: Confidential
⚠️ This is a recommended draft. Moiz to review and correct all
technical specifics before this is treated as operational.
What this workflow does: Governs the ongoing monitoring, patching, cost
management, and maintenance of all PrivateAI infrastructure — both internal
tools (n8n, Wiki.js, databases) and per-client environments — from the moment
they go live through to decommission.
What triggers it: Always active — runs continuously for every live
environment. Triggered additionally by: an alert from the monitoring system,
a reported incident, a scheduled maintenance window, or a cost review.
What it produces: Maintained, secure, and performant infrastructure across
all environments — with incident responses logged, patches applied on schedule,
and costs tracked against budget.
Why it matters: Monthly maintenance fees are PrivateAI's primary recurring
revenue. Clients pay for uptime, security, and managed infrastructure. An
unmonitored, unpatched environment is a liability that erodes that recurring
revenue when something goes wrong. INF-01b is what makes the recurring fee
defensible.
| Role |
Department |
Responsibility |
CEO? |
| DevOps Lead (Moiz) |
D5 |
Primary owner — monitors all environments, applies patches, manages costs, responds to infrastructure incidents |
No |
| AI Engineers |
D4 |
First to notice application-level issues that may have infrastructure root causes — report via Jira |
No |
| HoD |
D4 |
Escalation for infrastructure issues that affect delivery timelines or client commitments |
No |
| PM |
D4 |
Coordinates client communication if maintenance causes downtime or access issues |
No |
¶ 3. Maintenance cadence
| Activity |
Frequency |
Owner |
| Infrastructure health check |
Daily |
Moiz (automated monitoring) |
| Security patch review |
Weekly |
Moiz |
| Cost review (per environment) |
Monthly |
Moiz |
| Full environment audit |
Quarterly |
Moiz + HoD |
| SSL certificate renewal check |
Monthly |
Moiz |
| Backup verification |
Weekly |
Moiz |
| Access control review |
Monthly |
Moiz (cross-reference with HR records) |
Step 1 — Automated monitoring 🤖
- Who: Monitoring tool (TBD — pending OPS-O02)
- Action: Continuous monitoring of all live environments — uptime, CPU/memory
utilisation, error rates, network health, disk usage. Alerts sent to Moiz via
Discord when thresholds are breached.
- Tool: Monitoring tool (TBD — pending OPS-O02) · Discord (alerts) · AWS
CloudWatch (interim)
- Output: Real-time health visibility across all environments · alerts on
threshold breaches
Step 2 — Daily health check
- Who: Moiz
- Action: Review monitoring dashboard daily. Confirm all internal tools
(n8n, Wiki.js) and client environments are healthy. Log any anomalies in
Jira even if not yet actionable.
- Tool: Monitoring dashboard · Jira
- Output: Daily health status confirmed · anomalies logged
Step 3 — Infrastructure incident identified
- Who: Moiz (monitoring) or AI Engineer (reports)
- Action: Any infrastructure incident — downtime, performance degradation,
failed deployment, security alert — is logged in Jira immediately. Severity
classified:
- P1: Client-facing downtime or data access issue — immediate response
- P2: Degraded performance or non-critical service failure — response
within 2 hours
- P3: Internal tool issue, no client impact — response within 24 hours
- Tool: Jira (incident log) · Discord (team notification)
- Output: Incident logged with severity · response initiated
Step 4 — Incident response and resolution
- Who: Moiz (leads) · HoD (P1 escalation) · PM (client comms if needed)
- Action: Investigate root cause and resolve. P1 incidents: HoD and PM
notified immediately — PM coordinates client communication per DEL-06 process.
All incidents logged in Jira with: root cause, resolution steps, and preventive
action.
- Tool: AWS · n8n · Jira · DEL-06-DOC-01 (for P1 — client-impacting)
- Output: Incident resolved · root cause documented · preventive action
identified
- Decision: P1 client-impacting → DEL-06 triggered simultaneously
¶ Scheduled maintenance
Step 5 — Security patching
- Who: Moiz
- Action: Weekly review of available security patches for all infrastructure
components — OS, n8n, Wiki.js, AWS services, dependencies. Apply critical
patches within 48 hours of release. Apply non-critical patches in the next
scheduled maintenance window. Log all patches applied in Jira.
- Tool: AWS · n8n (version management) · Jira (patch log)
- Output: All environments patched and current
- SLA: Critical patches applied within 48 hours
Step 6 — Backup verification
- Who: Moiz
- Action: Weekly verification that backups for all critical systems are
completing successfully and restoration is possible. Test restoration quarterly.
Log backup status in Jira.
- Tool: AWS Backup · Jira
- Output: Backup health confirmed · restoration test results logged quarterly
Step 7 — Cost review
- Who: Moiz (reviews) · Finance Lead (informed)
- Action: Monthly review of AWS costs per environment. Compare against the
maintenance fee charged to each client. Flag any environment where costs are
approaching or exceeding the monthly fee. Notify HoD and Finance Lead of any
cost overruns.
- Tool: AWS Cost Explorer · Google Workspace (cost report) · Xero (revenue
reference)
- Output: Monthly cost report per environment · overruns flagged
Step 8 — Quarterly infrastructure audit
- Who: Moiz + HoD
- Action: Full quarterly audit of all environments — access control review
(cross-reference with HR active roster), unused resources identified and
decommissioned, architecture review against current security standards,
compliance posture check (encryption, audit logging, data retention). Findings
logged in Jira. Significant findings reviewed at L10 (OPS-O04a).
- Tool: AWS · Jira · Google Workspace (audit report)
- Output: Quarterly audit report · access list validated ·
unused resources decommissioned
Step 9 — Environment decommission (on engagement close)
- Who: Moiz (executes) · PM (requests) · Legal Lead (data retention check)
- Action: When an engagement closes (CS-06 Offboarding), PM requests
environment decommission from Moiz. Legal Lead confirms data retention
requirements (per INF-03 — default 90 days post-engagement). Moiz:
- Revokes all access (coordinated with OPS-H01d if contractor involved)
- Exports and archives any data per retention policy
- Decommissions AWS resources
- Documents decommission in the client infrastructure record
- Tool: AWS · Jira · Google Drive (archive)
- Output: Environment decommissioned · data retained or deleted per policy ·
infrastructure record updated
| Tool |
How it's used |
Agent-ready |
| AWS |
All environments — monitoring, patching, cost management |
Yes — confirmed |
| AWS CloudWatch |
Interim monitoring (confirmed monitoring tool TBD via OPS-O02) |
Yes — API |
| AWS Cost Explorer |
Monthly cost review per environment |
Yes — API |
| AWS Backup |
Backup and restoration |
Yes — API |
| n8n |
Internal automation platform — maintained here |
Yes — native |
| Wiki.js |
Internal knowledge base — maintained here |
Yes — self-hosted |
| Jira |
Incident log · patch log · audit findings · decommission log |
Yes — confirmed |
| Google Workspace |
Cost reports · audit reports · infrastructure records |
Yes — API + MCP |
| Discord |
Monitoring alerts · incident notifications · team comms |
Yes — API |
| Monitoring tool |
TBD — pending OPS-O02 |
TBD |
| Metric |
Type |
Target |
| Infrastructure uptime — all client environments (%) |
Leading |
≥ 99.5% |
| Critical security patches applied within 48 hours (%) |
Process |
100% |
| Weekly backup verification completed (%) |
Process |
100% |
| P1 incidents per month (count) |
Lagging |
0 |
| Monthly cost review completed (%) |
Process |
100% |
| Environments decommissioned within 5 days of engagement close (%) |
Process |
100% |
| Quarterly audit completed on schedule (%) |
Process |
100% |
Upstream:
Downstream:
| Version |
Date |
Description |
Author |
| 0.1 |
2026-07-18 |
Recommended draft — daily monitoring, incident severity (P1/P2/P3), patching SLA (48h critical), weekly backup verification, monthly cost review, quarterly audit, environment decommission process. Pending Moiz review. |
BA (Hammad Akram) |
← Back to Infrastructure Hub · ← Back to Company Operating System