# AI Deployment Brief — FDE Portfolio Worksheet

Original worksheet by Fouad Salkini. Not an official vendor curriculum,
credential application, hiring promise or executed deployment.

## Problem and users
- Workflow owner:
- Actual users:
- Problem / delay / failure to address:
- Current baseline and how it was measured:
- Why a model is appropriate (and which steps remain deterministic):

## Scope and acceptance
- In scope:
- Out of scope:
- Measurable acceptance targets agreed with the workflow owner:
- Expected behavior when evidence is absent or conflicting:
- Human approval points:

## Data and architecture
- Authorized data sources and their owners:
- Permissions and identity mapping:
- Data flows and external processors:
- Retention / deletion / audit requirements:
- Integration interfaces and tool capabilities:
- Service boundaries and secrets handling:

## Evaluation
- Representative normal cases:
- No-answer and stale-source cases:
- Cross-user / cross-project permission cases:
- Malicious source-content cases:
- Failure and recovery cases:
- Dataset split and protection against tuning to the evaluation:
- Expected and observed outputs:
- Actual latency, cost and human-review burden (unknown if unmeasured):

## Rollout and operations
- Release owner and approval:
- Limited rollout / feature flag:
- Monitoring and alert thresholds:
- Failure fallback and rollback:
- Incident contact and escalation:
- Known limits and unresolved findings:

## Adoption and handover
- User instructions and training:
- Maintenance owner:
- Runbook and dependency/model version record:
- Setup followed by another person:
- Handover issues found and fixed:
- Post-launch review date and outcome measures:

## Evidence record
- Repository / revision:
- Commands and checks actually run:
- Reports and observed results:
- Checks not run and why:
- Changes after verification: