Agentic AI Engineering Bootcamp
A hands-on, COMCEC-funded program that takes working developers from prompting to production-grade AI agents.
- Status
- Ongoing
- Years
- 2026–present
- Category
- Skills & employability
- Funded by
- COMCEC Al-Quds Program
12+
reusable agent skills published by the cohort
2
certification pathways (AZ-900 and AWS Cloud Practitioner)
About this project
Participants do not watch lectures about AI. They build agents, break them, review each other's code and ship reusable tools that anyone can install.
The program runs in person in Jerusalem and live online. It moves from LLM and reasoning foundations through core agent design patterns, cloud fundamentals, evaluation and judgement, and tools and APIs with the Model Context Protocol (MCP), to packaging knowledge as agent skills.
Work is assessed on evidence, not assertion. Every project is executed against live test cases and reviewed in writing, and every published skill has to prove that it activates when it should and stays quiet when it should not.
Participants can also follow an optional pathway to industry certifications: Microsoft Azure Fundamentals (AZ-900) and AWS Certified Cloud Practitioner.
Program journey
Module 0
Step 1: LLM and reasoning primer
How language models actually decide: tokens, context, sampling and the limits of reasoning.
Module 1
Step 2: Foundations and mental models
Prompt structure, instructions and context discipline, and which parts of a model's behavior you control.
Module 2
Step 3: Core agent design patterns
Control flow, tool use and orchestration, building the same agent in two frameworks to compare them.
Module 2.5
Step 4: Cloud fundamentals
The infrastructure agents run on, aligned to Azure Fundamentals, with a dedicated preparation intensive.
Module 3
Step 5: Evaluation and judgement
Building an agent that decides under constraint, with a hard budget on expensive calls.
Module 5
Step 6: Tools, APIs and MCP
The Model Context Protocol end to end: agents that discover tools they did not write and report honestly when they cannot verify something.
Module 5B
Step 7: Agent skills
Packaging knowledge so an agent loads it only when needed, and proving it with real transcripts.
