Beyond prompts · Beyond demos

The real depth of job-ready AI.

The visible tip is easy to demonstrate. The submerged capabilities—data, engineering, evaluation, security, operations and business judgment—make AI useful, dependable and economically viable.

ImageFirm competency architecture

Nine domains. Demonstrable capability.

This is an editorial learning framework, not a formal certification or a universal hiring standard. Select your intended role, open each domain, and rate your evidence of competence. Your self-assessment is saved only in this browser.

9Competency domains
0 / 9Domains assessed
—Average evidence score / 3
Important distinction: Not every role needs model training, advanced mathematics or fine-tuning. Conversely, safety, privacy, evaluation and human accountability cannot be skipped merely because a workflow uses a no-code platform.
Choose your depth

Three credible pathways

Specialization determines which domains require advanced proficiency. All pathways share foundational literacy, evaluation and responsible use.

01 · AI Application Engineer

Designs and ships robust AI-enabled software.

  • APIs, data systems and software architecture
  • Retrieval, evaluations, deployment and monitoring
  • Secure integration and incident response

02 · AI Automation Specialist

Redesigns workflows and integrates systems safely.

  • Process mapping, tools, APIs and orchestration
  • Approvals, exception handling and audit trails
  • Cycle-time, accuracy and cost measurement

03 · AI Product & Business Leader

Turns AI opportunities into governed business value.

  • Use-case prioritization and baseline measurement
  • Human-centered adoption and change management
  • Risk ownership, ROI and vendor assessment
From learning to evidence

A portfolio that proves something

Build one narrow, end-to-end application rather than ten disconnected demos. Example: a document-grounded assistant with permission-aware retrieval, cited answers, a human escalation path, evaluation dataset, latency and cost dashboard, and a documented rollback procedure.

01Define a user problem and measurable baseline
02Build, test and document the solution
03Demonstrate outcomes, limitations and safeguards

Authoritative references

NIST AI Risk Management Framework ↗ · NIST Generative AI Profile ↗ · OpenAI developer documentation ↗ · OWASP LLM security guidance ↗ · Google ML Crash Course ↗

Sources support the underlying engineering, risk and evaluation concepts; the nine-domain grouping, ratings and suggested projects are ImageFirm's editorial synthesis. Resources may evolve.