OpenFactoryAcademy

Level 1Associate

Associate Software Factory Engineer

The foundations of a software factory. What a software factory is, how agents and people share the work, why context and independent checks matter, how to find the bottleneck, and how to measure results. No code required.

For: Engineers starting to work in a software factory, and product and engineering leads who work alongside them.

What you will learn

  1. 01

    What a software factory is

    7 exam questions
    • Code got cheap, trust did not
    • From tools to a production line
    • The parts of a factory: work, workers, checks, evidence
  2. 02

    Agents do the work, people keep authority

    7 exam questions
    • What agents are good and bad at
    • Where human judgment stays
    • Approval as an explicit step
  3. 03

    Context: your knowledge makes AI useful

    7 exam questions
    • Why models do not know your business
    • Standards, exceptions and decisions as context
    • Keeping your knowledge when models change
  4. 04

    Independent verification

    7 exam questions
    • One model makes, another checks
    • Why the checker must not be the maker
    • Tests, review and gates as layers
  5. 05

    Find the bottleneck, then the next one

    6 exam questions
    • Faster coding moves the queue
    • Reading where work waits
    • Fixing one constraint at a time
  6. 06

    Measuring results

    6 exam questions
    • Throughput, lead time, rework, human intervention
    • Measure production, not tokens or pull requests
    • Setting a baseline

The exam

Questions
40, multiple choice
Time
60 minutes
Pass mark
80%
Price
$25, one free retry
You earn
Certified Associate Software Factory Engineer

Sample questions

  1. Agents do the work, people keep authority

    Your team wants more autonomy on its line. Which change raises autonomy without deleting a safeguard?

    1. Remove the manual review step entirely, since any step that needs a person slows the line down.
    2. Let the maker agent review its own work instead of a person, since it knows the change best.
    3. Replace a manual review step with an independent check that records evidence.
    4. Keep the review step but let reviewers approve without reading, so that the queue clears faster.

  2. Independent verification

    A team uses one model to write a feature and the same model, in the same session, to review it. The review finds nothing, but customers soon report a bug. What change would most directly address the cause?

    1. Use a checker from another vendor that judges against the spec.
    2. Ask the same model to review the feature twice, with a stronger instruction to look for mistakes.
    3. Add more tests written by the maker, so that the review has more evidence to confirm the change.
    4. Remove the review step, since it caught nothing.

  3. Measuring results

    A leader asks whether the team's agents are paying off. Which answer measures production?

    1. Token usage rose sharply this quarter, which shows how much more work the agents are doing for us.
    2. Stories reaching production rose from the baseline, and lead time fell.
    3. Merged pull requests doubled this quarter.
    4. Every engineer now uses agents daily.

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

Next level: Professional Software Factory Engineer