Prompt Engineering Training

Description :

Discover the training program “Prompt Engineering Training”, an operational 1.5-day course designed to master prompt engineering techniques, understand the capabilities (and limitations) of large language models, and learn how to design AI agent–oriented use cases (tools, RAG, protocols, and best practices).

The program is structured in two phases: a first half-day (3h30) dedicated to conceptual understanding and trainer-led demonstrations (LLM fundamentals, prompting levels, advanced techniques, multimodality and deep research, introduction to AI agents), followed by a full day of hands-on workshops (7h) focused on practical application: writing robust prompts, structuring outputs, building a simple RAG pipeline, introduction to agentic logic, and connecting tools/servers (MCP), with guided exercises and feedback.

The objective is to enable participants to gain autonomy in their daily use of AI assistants, improve the quality and reliability of generated outputs, and understand how prompt engineering connects with RAG and agentic architectures to industrialize AI use cases in an enterprise context.

Learning objectives :

By the end of the training, participants will be able to:

• Explain key LLM concepts: tokens, context window, hallucinations, and high-level training principles (Transformers, RLHF).
• Apply core prompt engineering best practices (clarity, separation of instructions and context, specificity, examples) using an iterative improvement approach.
• Structure prompts using frameworks (role, steps, context, constraints, output format) to obtain more actionable outputs.
• Leverage advanced prompting techniques (few-shot, rephrase & respond, chain-of-thought, self-consistency, chain-of-verification) to improve output quality, robustness, and reliability.
• Understand the value of multimodality, deep research, and reasoning models, and identify relevant business use cases.
• Describe what an AI agent is (instructions, context, tools, execution loops, human-in-the-loop) and differentiate “function calling” from “agent orchestration” approaches.
• Explain the principles of a RAG pipeline (retrieval, context selection, generation) and its benefits and limitations (including “lost-in-the-middle”).
• Apply concepts through concrete use cases: output structuring, reusable prompt creation, building a mini-RAG, introduction to tool/server connections (MCP), and best practices for security and observability.

Target audience :

Business professionals, innovation teams, project managers, consultants, non-specialist IT/data profiles, and anyone looking to improve their interactions with LLMs and understand the fundamentals of agentic AI.

No prior AI expertise is required, though basic digital literacy is recommended.

Duration :

1.5 days: Half-day (3h30) + Full workshop day (7h) — total of 10.5 hours.

Number of participants :

Maximum 12 participants.

Prerequisites :

• Laptop with Internet access
• Basic computer skills
• Access to an LLM tool (e.g., ChatGPT, Claude, Gemini) recommended for workshops

Schedule :

Half-day 1 (3h30): Concepts & demonstrations

• Introduction: LLM fundamentals (tokens, context window, hallucinations, training principles: Transformers, RLHF)
• Prompting levels: basic / advanced / reasoning models / multimodality
• Prompt structuring frameworks (role, steps, context, constraints, output format)
• Advanced techniques: few-shot, rephrase & respond, chain-of-thought, tree-of-thoughts, self-consistency, chain-of-verification
• Multimodality & deep research: use cases (text2image, image2text, text2code), strengths and limits
• Introduction to AI agents: context, tools, frameworks, MCP; RAG principles
• Best practices: observability & security (LLMOps/AgentsOps, prompt CI/CD, evaluation)
• Break: 10 minutes

Day 2 (7h): Hands-on workshops (can be trailor-made to your needs if necessary)

• Workshop 1: prompt diagnostics and improvement (iteration, specificity, context vs instructions, constraints)
• Workshop 2: advanced prompting (few-shot, CoT, CoVe) on business cases: summarization, extraction, classification, structured writing
• Workshop 3: output structuring (tables, JSON, constrained formats) and comparison of generation strategies
• Workshop 4: building a mini-RAG (documents, retrieval, generation) and discussion of limitations (lost-in-the-middle)
• Workshop 5: introduction to agentic AI (tools, loops, human-in-the-loop), MCP connection demonstration
• Workshop 6: quality & security best practices (prompt testing, evaluation, guardrails, traceability)
• Wrap-up: recap, adoption roadmap, and Q&A

Pricing :

• Inter-company : upon request
• Intra-company : upon request

Teaching methods :

• Structured, accessible theoretical input
• Live demonstrations (prompting, strategy comparison, multimodality, deep research)
• Guided hands-on workshops with feedback
• Discussions and participant experience sharing

Assessment :

• Pre- and post-training self-assessment and final quiz
• Practical exercises with trainer feedback

Delivery format :

In-person or remote training (Microsoft Teams).
A computer with webcam and microphone is required for remote sessions.

Access timeline :

Registration possible up to 7 business days before the session.

Accessibility :

Training accessible to people with disabilities.
Contact: accessibility@eurekia-learning.com

Contacts :

• Quality referent : Jihane Khouzaimi – +33 6 86 05 04 02 – contact@eurekia-learning.com
• Educational referent : Hatim Khouzaimi