How we structure an engagement.

Five phases, each with a goal, concrete activities and outputs you keep. This is the structure we propose for new work: the depth of each phase is scaled to the problem, and a small engagement may cover several of them in a week.

1. 01

   Discovery & Assessment

   GoalUnderstand the system, the constraints and the real objective before proposing anything.

   Activities

   • Read-only review of architecture, cloud accounts, repositories and pipelines
   • Written questions to the people who run the system
   • A baseline of cost, reliability and delivery metrics, where they exist
   • A list of risks ranked by impact and likelihood

   Outputs

   • Assessment report with evidence
   • Prioritised list of problems and opportunities

   You decideWhether to go on, and with which priorities.

2. 02

   Architecture & Planning

   GoalChoose a design deliberately and turn it into a plan that can be carried out in small steps.

   Activities

   • Options compared on cost, risk, effort and operability
   • Target architecture and the path to reach it
   • Decision records for every significant choice
   • A sequenced plan with fixed-scope milestones

   Outputs

   • Architecture document and diagrams
   • Plan with scope, milestones and acceptance criteria

   You decideApproval of the design and of the first milestone.

3. 03

   Implementation

   GoalBuild the change incrementally, as code, in your repositories and accounts.

   Activities

   • Infrastructure, pipeline and software changes through reviewed pull requests
   • Staged rollouts or feature flags where the risk is high
   • Regular written progress against the plan
   • No manual change in production that is not captured in code

   Outputs

   • Reviewed, merged code
   • Documentation updated with each change

   You decideAcceptance of each milestone against its criteria.

4. 04

   Validation & Reliability

   GoalShow with evidence that the system behaves as designed — under load, under failure and during recovery.

   Activities

   • Load and failure tests sized to real traffic
   • Restores and rollbacks exercised, not assumed
   • SLOs, dashboards and alerts defined and tuned
   • Review of identities, exposure and secrets

   Outputs

   • Test results and known limits
   • SLOs, alert rules and dashboards as code

   You decideGo-live, with the remaining risks stated in writing.

5. 05

   Knowledge Transfer & Continuous Improvement

   GoalMake your team the owner of the result, then improve it with data.

   Activities

   • Runbooks for routine operations and incidents
   • Handover in writing, with working sessions if your team wants them
   • Review of SLOs, costs and incidents after go-live
   • A backlog of improvements ranked by measured impact

   Outputs

   • Runbooks and operational documentation
   • Improvement backlog with measured priorities

   You decideWhether to continue, and on what — or to run it on your own.