Master · by application only100 hours · 7 modules · one-to-one or small group

Master Executive in AI Agent Engineering

Designing agents that hold up in real work, not just in the demo

Building an agent that works once is easy; building one that handles thousands of unpredictable requests without surprising anyone is a craft. The programme covers tool design, context management, systematic evaluation and running in production.

What the programme gives you

  • Design an agent's tool set and the schemas of its parameters
  • Handle context and memory across long tasks without losing quality
  • Build evaluation sets that catch regressions before your users do
  • Keep cost and latency in check without giving up reliability
  • Defend the agent against prompt injection and tool misuse

Who it is for

  • Backend developers starting to build agents
  • Data scientists and ML engineers moving into agentic systems
  • Solution architects who have to weigh up feasibility and risk

Entry requirements

  • Backend development experience and hands-on use of APIs
  • No degree required
  • Familiarity with at least one server-side language

Selection is based on professional experience. No degree is required.

Module syllabus

A limited-places programme, one-to-one with a dedicated tutor or in a small group, with project work on an agent built from scratch.

  1. 01

    Anatomy of an agent

    • The observe-decide-act loop and when it should stop
    • The difference between rule-based automation, workflows and agents
    • When an agent is the wrong choice
  2. 02

    Tool design

    • Parameter schemas and descriptions the model reads the way you meant
    • Granularity: a few powerful tools or many narrow ones
    • Errors as information the agent can use, not as exceptions
  3. 03

    Context and memory

    • Retrieval strategies for the information that actually matters
    • Compacting context across long sessions
    • State that persists from one run to the next
  4. 04

    Systematic evaluation

    • Building test sets that represent the domain
    • Automated scoring alongside human spot checks
    • Telling serious failures apart from statistical noise
  5. 05

    Production

    • Keeping cost and latency under control at real load
    • Observability: what to trace so you can diagnose
    • Graceful degradation when a downstream service stops answering
  6. 06

    Security

    • Prompt injection from untrusted content
    • Permission boundaries and human confirmation on destructive actions
    • Handling personal data inside the conversation context
  7. 07

    Project work

    • Designing and building an agent for a concrete case
    • Building the evaluation set and measuring the results
    • Final presentation and a critical review of the architecture

Format

How you follow the programme

There are two ways to take the Master. The syllabus, the hours and the credential are the same either way: what changes is how many people you do it with.

  • Private

    One-to-one with a dedicated tutor

    The programme is yours alone: a tutor who follows you and nobody else, a schedule agreed around your commitments, and project work built on the job you actually do.

    • Pace and times agreed together, not set by a classroom timetable
    • Every session starts from your own cases, not from generic examples
    • Code and project reviewed line by line
  • Small group

    A few people on the same programme

    A handful of people working through the same syllabus together. Comparing notes with peers adds something, and the group stays small enough to leave room for everyone.

    • People with different backgrounds discussing the same problems
    • Guaranteed room for everyone's questions
    • Project work still individual

Either way, access remains by selection: every request is reviewed individually. Tell us which format you prefer in your application, and we will talk it through at the interview.

Executive Master

The credential you receive

On completing the programme and defending your project work you receive a Certificate of Completion, with a permanent public verification page and a downloadable PDF. It is an Open Badge compliant with the 2.0 standard: not an image, but a credential anyone can verify automatically.

  • Open Badge 2.0

    An open standard: readable by skill wallets and by LinkedIn.

  • Official PDF

    A document generated from the database, with a unique credential code.

  • Impossible to fake

    The code is unique and verification reads from our own database.

Advanced professional training programme delivered by Rinoova. It is not an academic qualification nor a university master’s degree under Italian Ministerial Decree 270/2004.

Access

How selection works

Places are limited and every request is reviewed individually, for the private format as much as for the small group. It is not a formality: we look at whether the Master is useful to you right now, and if it is not we say so.

  1. 01

    Application

    You write and introduce yourself: your role, your experience, why the programme interests you and how you would prefer to follow it.

  2. 02

    Review

    We read every application and always reply, including when the answer is no.

  3. 03

    Interview

    If your profile fits, half an hour to talk through goals and what each side expects.

  4. 04

    Admission

    If you are admitted, you receive the detailed syllabus, the fees and the dates of the next intake.

Apply for the programme

The Master has limited places and access is by selection. Send us your application: we read it and reply either way, including when the answer is no.

  • Access by selection: every request is reviewed individually
  • Privately with a dedicated tutor, or in a small group
  • Individual project work reviewed by a teacher
Write to us

Email address

academy@rinoova.com

Open your email app

What we need to know

  • Full name:
  • Current role:
  • Years of experience:
  • Why you want to take part (a few concrete lines):
  • Preferred format (one-to-one with a dedicated tutor / small group):
  • GitHub / LinkedIn (optional):

The button pre-fills the subject and the message: you can edit both before sending. We collect no data on this site, the email comes straight to us.