James KamugishaWe are a blip in time.
How I build

From validated need to a product that holds up in the field.

The product engineering process I run for software, hardware, and hybrid systems. It is evidence gated and iterative. Every stage has exit criteria, every decision is recorded, and any result can send the work back. That is by design.

Open OpenRelay Agents run this process on a problem. Invite only for now.
field data becomesthe next problem brief?1Problem2Investigate3Define4Prototype5Test6Build7Operatea decision gate, detailed in the flowchart
The process

Seven stages, one loop

Work moves clockwise, and the last stage feeds the first. The orange diamonds mark the two decision gates where the work can change direction. Select a stage for its purpose, activities, a worked example, and its exit criteria.

Iteration

Where the work turns back

The loop shows the lifecycle. This shows the control flow: every path a gate decision can send the work down.

Choose a turn to trace its path in the chart, or a stage in the chart to read about it.

An observed needyesmetrevise the designnew baselinerestate the problemnot metfield data becomes the next problem briefUnmet needconfirmed?noStopneed already servedCriteriamet?no goStopno viable caseRetireat end of life?1ProblemIs the need real?2InvestigateWhere is the gap?3DefineWhat must it achieve?4PrototypeWhat is the top risk?5TestDoes it meet spec?6BuildIs it repeatable?7OperateDoes it hold in use?forward patha gate sends work backthe work ends

Swipe sideways to see the whole chart.

One process

Three product types

Software

The configuration item is a tagged commit, a dependency lock, and an environment configuration. Reproduction is a deterministic build.

Physical

The configuration item is a drawing revision, a bill of materials, and the serialized unit built from them.

Hybrid

Hardware, firmware, and app are verified as one named configuration. Change one element, and every dependent test is flagged for regression.

Work with me

Have a problem worth engineering?

Bring the user, the context of use, and the failure you keep seeing. Discovery starts there.