Learn everything from core agent design to advanced multi-agent orchestration. In just 8 weeks, build the advanced skills needed for specialized AI engineering roles.
You will learn to design stateful workflows, orchestrate multi-agent systems, and build with modern stack standards including the OpenAI Agents SDK, LangGraph, Claude Agent SDK, Pydantic AI, LlamaIndex, Modal, Langfuse, and the Model Context Protocol (MCP). The curriculum covers bridging prototype agents into shippable autonomous systems through rigorous production engineering, complete observability, and full instrumentation. Through morning theory and afternoon coding sessions, you will build weekly projects that lead into a portfolio-ready capstone.
What's in store at our Agentic AI bootcamp
Master the foundations of agentic AI by building a hand-rolled ReAct loop in pure Python and refactoring it with the OpenAI Agents SDK. You’ll gain practical experience executing tools safely in Modal sandboxes and setting up full observability with Langfuse logging. For your weekly project, you'll build and track an autonomous research assistant agent that executes multi-step web searches and synthesizes cited findings.
Master LangGraph to build durable, stateful AI workflows featuring conditional routing, event-level streaming, and human-in-the-loop approval gates with idempotency. You will gain hands-on experience instrumenting complete end-to-end observability by setting up self-hosted Langfuse tracing. For your weekly project, you’ll develop and monitor a production-ready, multi-step agent equipped with state checkpointing and human control limits.
Learn to equip AI agents with persistent knowledge and long-term memory by mastering agentic RAG, adaptive query routing, and advanced memory architectures—from short-term context to knowledge graphs. Using LlamaIndex, you will transform retrieval into an active tool that allows agents to critique and refine their own queries. For your weekly project, you’ll build a domain-specific Knowledge Copilot that delivers citation-grounded answers and maintains memory across user sessions.
Expand your toolkit beyond LangGraph by mastering Model Context Protocol (MCP) servers, autonomous tool-use via the Claude Agent SDK, and type-safe development with Pydantic AI. You will learn to build, authenticate, and integrate custom MCP servers to connect agents seamlessly with external systems. For your weekly project, you’ll construct an end-to-end workflow automation agent that orchestrates tools across platforms like GitHub, Slack, Notion, and your own custom MCP server.
Develop multi-agent orchestration patterns—like Orchestrator-Worker, Supervisor, and Swarm—while optimizing systems for cost, latency, production safety, and EU AI Act compliance. You will implement rigorous eval-driven development using DeepEval and guard your applications against OWASP top vulnerabilities with a multi-layered architecture. For your weekly project, you’ll productionize and deploy a previously built agent to Modal, complete with an evaluation suite, guardrails, and self-hosted tracing.
Frame your agentic AI capstone by selecting a real-world archetype, defining clear success metrics, and designing the target architecture. You will set up production-ready sandboxes and observability infrastructure to deliver an initial, end-to-end working implementation.
Build your production agent loop by integrating essential tools, knowledge retrieval, and memory architectures. You will then refine system performance through evaluation-driven iteration and feedback from mid-capstone peer and coach presentations.
Make your final project production-ready using strong multi-layered security, complete monitoring dashboards, and clear documentation. You will deploy your application to production and showcase your fully instrumented, autonomous agent in a live graduation presentation.
The most up-to-date, effective tech stack on the market:
Implement custom ReAct loops in pure Python before scaling with OpenAI Agents SDK and Claude Agent SDK.
Build durable workflows in LangGraph with state checkpointing, conditional routing, and human-in-the-loop approval gates.
Deploy agentic RAG and adaptive query routing using LlamaIndex across vector stores, episodic memory, and graph memory.
Publish custom Model Context Protocol (MCP) servers and enforce type-safe contract validation using Pydantic AI.
Orchestrate multi-agent topologies—including Supervisor, Orchestrator-Worker, and Swarm patterns—optimized for cost and latency.
Instrument end-to-end tracing in Langfuse, automate evals with DeepEval, and enforce EU AI Act safety controls on Modal.
Our career service helps you turn your new skills into your next role with CV reviews, LinkedIn feedback, and interview coaching.
Spicedlings are getting hired by your favourite companies:
Invest in your future
If you’re registered as unemployed (or soon to be) in Germany, you could be eligible to have all your costs covered with a Bildungsgutschein (training voucher).
For more information on this option, check our page dedicated to financing your coding bootcamp with a Bildungsgutschein.
We want to make our best-in-class tech courses available to everyone with the motivation to complete them.
Our Deferred Payment Option enables those who aren’t in the position to pay upfront nor in instalments to participate, by offering the chance to pay back at a later date.
If you’re ready to cover the cost of our coding bootcamps immediately, this is the option for you. Pay 14 days before the course starts.
8 Weeks | Full-time
The Agentic AI training course is aimed at software developers and tech professionals who want to take the direct step toward becoming AI engineers and build production-ready, autonomous AI systems.
The program runs Monday to Friday, 9:00–18:00 with 45 minutes lunch break. This is a full-time commitment equivalent in time to a full-time job. You should expect an immersive, fast-paced learning curve that balances theory and daily hands-on application. Every day combines live coach-led lessons with guided practical work under coach and TA guidance.
Upon completing the Agentic AI training, you will possess a robust profile combining technical expertise, practical development skills, and project-based know-how. The skills acquired break down as follows:
1. Technical & Development Skills
Development of autonomous AI agents: You will learn to build agentic AI systems from scratch—ranging from simple ReAct loops (in Python) to refactoring and structuring using the OpenAI Agents SDK.
Integration of LLMs & tool-calling: You will gain the ability to integrate Large Language Models (LLMs) and connect them with tools (tool-calling), interfaces, and databases.
Secure execution & sandboxing: You will learn how to securely execute external tools and AI code within isolated environments (such as Modal sandboxes).
Observability & logging: You will be able to comprehensively monitor, analyze, and debug AI agent behavior (e.g., using Langfuse logging).
Multi-step & autonomous processes: You will learn to build multi-stage, autonomous workflows—such as a research assistant that independently searches the web, filters information, and summarizes sources in a structured manner.
2. Methodological & Practical Skills
Hands-on prototyping: You will go beyond theory to implement concrete, practical projects every week.
Problem-solving & debugging: You will acquire the ability to systematically analyze unpredictable behavior in language models and AI agents, troubleshoot issues, and optimize output quality.
Project & portfolio work: Through the final project, you will demonstrate your ability to independently develop complex AI solutions, taking them from the initial idea to a fully functional application.
3. Role-relevant soft skills & working methods
Agile work in remote teams: You gain experience in collaborative software development within distributed teams.
English as the working language: Since the course is taught in English, you acquire the necessary language and technical proficiency to work in international tech teams and with global open-source standards.
Yes, absolutely. If you are currently unemployed, facing a potential layoff, or looking for a career change, the Agentur für Arbeit or Jobcenter can cover 100% of the bootcamp costs through an Education Voucher (Bildungsgutschein).
The process involves three steps: registering as a jobseeker with your local Agentur für Arbeit or Jobcenter, requesting our official training documentation (we provide this within one working day), and submitting the voucher application. Approval typically takes two to six weeks — cohort sizes are limited to approximately 15 participants, so starting early matters.
Additionally, we offer several alternative financing paths if you don't qualify for government funding, for example: Self-Payment & Installment Plans, Scholarships, Employer Sponsorship. Please find more details on our financing page.
You can join from anywhere in Germany! No matter whether you live in Berlin or Hamburg, Cologne or Düsseldorf, Hanover or Bremen, Leipzig or Dresden, or anywhere else, you can complete this bootcamp entirely online.
Join our growing lively community and fast track your future to a career in tech
Email address
I would like to receive email updates from SPICED Academy. This decision can be revoked at any time. Information on how we handle your data can be found in our privacy policy.
