OpenFactoryAcademy

Level 2Professional

Professional Software Factory Engineer

Building a software factory. How to set up a software factory on a real workflow: baseline first, then the line, the agent roles and their context, the checks, rework and exceptions, the evidence ledger, and a pilot on real work.

For: Engineers, forward deployed engineers and partner delivery teams who set up a factory. Associate level, or the same experience, recommended first.

What you will learn

  1. 01

    Pick the workflow, set the baseline

    8 exam questions
    • Choosing a workflow that matters
    • Tracing it from approved request to production
    • Recording the baseline before changing anything
  2. 02

    Design the line

    7 exam questions
    • Spec, build, verify, ship as stations
    • Right-sized units of work
    • Hand-offs between stations
  3. 03

    Agent roles and context

    7 exam questions
    • Builder, reviewer and operator roles
    • What each role sees and must not see
    • Mixing vendors on purpose
  4. 04

    Designing the checks

    7 exam questions
    • Tests that prove the right thing
    • Independent model review
    • Approval gates and who owns them
  5. 05

    Rework and exceptions

    7 exam questions
    • Sending work back with a reason
    • Routing exceptions to people
    • Turning repeated corrections into defaults
  6. 06

    The evidence ledger

    7 exam questions
    • Recording every assignment, check and decision
    • Append-only by design
    • Using the record in review and audit
  7. 07

    Pilot on real work

    7 exam questions
    • Running the line next to the team
    • Comparing against the baseline
    • Deciding what to change next

The exam

Questions
50, multiple choice
Time
75 minutes
Pass mark
80%
Price
$25, one free retry
You earn
Certified Professional Software Factory Engineer

Sample questions

  1. Pick the workflow, set the baseline

    You are setting up a factory for a team whose changes take weeks to reach production. The team lead wants to start building agents this week and measure later, once the line is running. What should you do first?

    1. Record the baseline, so later results can be compared with the state before any change.
    2. Start building agents now, since the baseline can be reconstructed afterward from the agents' own logs.
    3. Measure tokens and pull requests per developer first.
    4. Pick the bottleneck from experience and design around it.

  2. Agent roles and context

    A team gives its reviewer agent the spec, the diff, the test results, the standards and the builder's notes explaining its reasoning. Reviews now pass almost everything. Which input should be removed first?

    1. The builder's notes explaining its reasoning.
    2. The test results, so it judges the code alone.
    3. The standards, since they make the reviewer too lenient on matters of style.
    4. The spec, since the diff shows the intent.

  3. Rework and exceptions

    A reviewer agent sends a unit back three times because its retry behavior does not match the spec. Each time, the builder changes it in a different way. A person reads the spec and finds it says only 'retry on failure'. Where should the unit go next?

    1. Back to build, with a stronger builder.
    2. Forward to ship, after three tries.
    3. To the reviewer, which should write the retry policy into the spec itself.
    4. Back to spec, so a clear retry policy is written.

Sign in for free practice questions, each with the answer explained.