Mandragore Learning Academy
DE EN ES FR IT NL PT PT-BR
Careers & jobs

AI Engineer: role, skills and learning paths for production AI

Explore the AI Engineer role: APIs, RAG, agents, evaluation, security, deployment and Mandragore Learning Academy learning paths.

Who is this article for?

For developers, cloud engineers, data professionals and technical profiles who want to move from using AI tools to engineering reliable AI-enabled systems.

What you need to understand

AI Engineering sits at the intersection of software development, data, cloud and AI.

A production solution must be integrable, testable, secure and observable, with explicit handling of failure modes, costs and limitations.

Why this skill matters today

Generative AI is moving from prototypes into products and workflows.

Organisations therefore need people who can connect models to existing systems while managing quality, security, data access and governance.

What you actually do with this skill

Integrate modelsCall APIs and handle inputs, outputs and failures.
Build agentsAllow controlled tool use and actions.
Implement RAGGround answers in authorised sources.
Evaluate qualityCreate test sets and acceptance criteria.
Operate in productionManage secrets, cost, latency, logs and incidents.

Concrete example: Internal knowledge assistant

A company wants support teams to query internal documents without exposing restricted information.

Problem : Answer quickly while using controlled sources.

Context : Internal documents, differentiated permissions and citation needs.

Skills : APIs, RAG, authentication, context, evaluation and monitoring.

Example : Retrieve authorised passages and generate a sourced answer.

Hands-on application : Test ambiguous cases, access controls and prompt injection before deployment.

Skills to master

Professional mastery develops progressively from foundations to autonomous practice.

BeginnerPython, REST APIs, generative AI, JSON and cloud basics.
IntermediateRAG, agents, tools, vector search, testing and security.
Advanced / autonomousArchitecture, evaluation, observability, deployment, cost and governance.

Errors, limitations and points of attention

  • Trusting outputs without verification.
  • Exposing secrets in prompts.
  • Giving agents excessive permissions.
  • Ignoring cost and quotas.
  • Deploying without diagnostic traces.

How to progress

BeginnerPython, REST APIs, generative AI, JSON and cloud basics.
IntermediateRAG, agents, tools, vector search, testing and security.
Advanced / autonomousArchitecture, evaluation, observability, deployment, cost and governance.
Mandragore Learning Academy

Train with Mandragore Learning Academy

Mandragore Learning Academy does not yet offer one professional programme named exactly “AI Engineer”. The AI Engineer competency path instead connects the existing training that is genuinely relevant to the role.

Relevant building blocks include Machine Learning, AI Agents, Python, API & Web Services, Cloud and AI-assisted development.

View recommended training →
Related competency path

Related competency path

The related competency path shows essential, recommended and complementary skills around this role.

View the competency path →
Sources and references

Mandragore source: config_metiers.php remains the source of truth for the role and learning paths actually available.

  • Microsoft Learn — AI-103 — Developing AI Apps and Agents on Azure
  • OECD — AI and skills: What we know so far