Let’s talk
EnglishEspañol
Let’s talk

AI Deployment Engineer at OpenAI

AI training and advisory for teamsFrom promptto production.

I help teams learn to work with AI and make the technical decisions that turn a promising demo into a system people can operate every day.

Plan team trainingRequest an AI review

Where I’ve worked

  • OpenAI
  • GitLab
  • n8n
  • Splunk

The path

An AI demo takes an afternoon. A system people can operate takes decisions at every step.

These are the five stages I help teams work through. Training covers the first two; advisory focuses on the last three.

  1. 01Prompt

    Understand what the model can do

    Your team learns how models behave, where they fail and how to write instructions that can be tested and improved.

    Training

  2. 02Prototype

    Turn a task into a workflow

    We take a real task from your work and build a first version with the tools the team already uses.

    Training

  3. 03Tools and agents

    Connect it to your systems

    Agents, APIs and automations: which decisions belong to the model and which ones code should control.

    Advisory

  4. 04Evaluation

    Measure before you trust

    Define what a good result looks like, test it against real cases and decide where a person needs to review.

    Advisory

  5. 05Production

    Operate it every day

    Costs, monitoring, security and ownership: the decisions that keep a system useful after launch.

    Advisory

Diagnosis

Where is your project on the path?

Five questions, one per stage. It is an informal self-assessment: nothing is stored, and it does not replace a conversation.

  1. Prompt. Does your team know where the model fails on its task, and how to check it?
  2. Prototype. Is a real task already running as a first workflow with the team’s tools?
  3. Tools and agents. Is it clear which decisions the model makes and which ones code controls?
  4. Evaluation. Are results measured against real cases before anyone relies on them?
  5. Production. Are owners, costs and monitoring defined for running it every day?

Your path

0 of 5 answered

Prompt: no answer. Prototype: no answer. Tools and agents: no answer. Evaluation: no answer. Production: no answer0102030405

Answer the five questions to see your next step.

Two ways to work together

Build the skills. Review the decisions.

For leadership, technical and mixed teams

Team training

Practical workshops built around your team’s workflows, tools and starting point.

Stages

  • 01 Prompt
  • 02 Prototype

Example deliverable

A workflow the team can explain, test and adapt, with materials to keep practising.

Plan your trainingHow the training works

Before you build or scale an AI system

AI advisory

A focused review of architecture, agent workflows and operational risks before you commit to an approach.

Stages

  • 03 Tools and agents
  • 04 Evaluation
  • 05 Production

Example deliverable

A written review with prioritised recommendations and the decisions to make next.

Discuss an AI reviewHow the advisory works
Idir Ouhab portrait

Idir Ouhab

Behind the work

Engineer, trainer and speaker.

Across GitLab, Splunk, n8n and now OpenAI, I have worked in solution design, technical sales and enablement. I teach on Platzi, host Prompt&Play and speak about AI and automation in English and Spanish.

Invite me to speakRead my story
  • Idir Ouhab speaks with a microphone on stage at OpenAI Academy, in collaboration with Digitaliza Madrid.
    Idir Ouhab speaking at OpenAI Academy, in collaboration with Digitaliza Madrid.
  • Idir Ouhab gives a talk on AI systems to the audience at AI Tech Summit in Belgrade.
    Idir Ouhab at AI Tech Summit, Belgrade.
  • Idir Ouhab and four other speakers take part in a panel discussion before the Tech Show Madrid audience.
    Panel discussion on AI at Tech Show Madrid.
  • Participants in the Dify, cognee, and N26 hackathon gathered for a group photo.
    Dify, cognee, and N26 hackathon. I helped teams with their prototypes and served as a judge.

Learn by playing

AI makes more sense when you try it.

“Help Lucía” is a free game I created to learn how to use AI from scratch. Lucía runs a bakery and you help her make the most of AI: asking well, spotting mistakes, getting ideas and avoiding scams.

missions
8
places
8
rules
8
Start playing

No sign-up. Your progress stays in this browser.

Lucía smiling with her phone outside the bakery, with Marta, Sara and Andrés

Blog

AI in practice

Browse the blog
12 min read

AI harnesses: the missing layer between a smart model and useful work

Read the article

If a website says Jev means “goodbye to LLMs,” close the damn tab

Late night shows, 1,200 AIs and OpenAI's rooms

Get new articles by email
Podcast in Spanish

Prompt&Play

A podcast about AI and automation.

Latest episode

Una imagen, Claude y un repositorio privado de OpenAI | #31

See all episodes

Contact

What are you working on?

Tell me what your team needs. I’ll reply by email to discuss the scope and availability.

Choose the service that best fits your request.

I’ll use your details only to respond to your enquiry. Privacy policy

Cookie preferences

Necessary storage stays active. Choose each optional purpose below; closing this panel does not change your saved choices.

Necessary storage

Always active

Remembers your language and cookie choices. A separate session cookie keeps administrators signed in. These are not used for advertising.

Allows article-view counts and Google Analytics, provided by Google, to measure visits and interactions such as inquiries. Google receives connection data and uses measurement cookies. Form contents and email addresses are excluded; advertising features are disabled.

Remembers your permission for external audio and embedded content. Each item still waits until you choose to play or load it, and identifies its provider first. Providers may receive your IP address and set their own cookies. You can also allow a single item without saving a general preference.

Your choices are valid for 180 days in this browser. Optional purposes start switched off. If browser storage is unavailable, your choice lasts for this page only.

You can withdraw permission here at any time. If optional content has already loaded, the page reloads to stop it; unsent form changes may be lost.

How cookies and storage are used