Mastering OpenAI Codex: Agentic Software Engineering

Available on Request

Mastering OpenAI Codex: Agentic Software Engineering

Learn Codex end-to-end, from LLM foundations to MCP, BMAD, and spec-driven agentic delivery.

Duration

3 days

Schedule

Flexible Dates

Available as in-house or custom-scheduled workshop

Your Trainers

Björn Wilmsmann

he/him

AI-based automation, AI-augmented software engineering, Spring, Java, web technologies

Independent IT consultant, trainer, and published author with deep expertise in Java, Spring Boot, and cloud-native architectures. As co-author of "Stratospheric: From Zero to Production with Spring Boot and AWS," Björn has helped thousands of developers navigate the path from local development to production-ready cloud deployments. He specializes in digital transformation, helping businesses automate processes, improve software quality, and make critical data accessible through well-designed APIs. His client work spans enterprise application development, architecture and design reviews, Angular and Spring Boot training, and hands-on coaching of development teams across industries. Björn brings a pragmatic, business-focused approach to technology. Whether advising on architecture decisions, conducting software quality audits, or running customized workshop formats, he focuses on delivering measurable impact and enabling teams to work independently with new technologies.

View Profile

Pawel Sawicki

he/him

Senior Machine Learning Engineer

Machine Learning Engineer and Software Architect with 15+ years of experience, specializing in deep learning, NLP, and LLM-based systems. Strong focus on designing and deploying end-to-end AI pipelines, including model fine-tuning (LoRA/PEFT), retrieval-augmented generation (RAG), and high-performance inference with modern frameworks such as PyTorch, Hugging Face Transformers, and vLLM. Experienced in building agentic systems using LangChain and LangGraph, as well as developing scalable backend services in Python. Combines solid software engineering principles with hands-on expertise in data-centric AI to deliver robust, production-ready solutions.

View Profile

About This Workshop

A 3-day hands-on workshop that takes participants from the fundamentals of large language models to advanced agentic development workflows with OpenAI Codex. The workshop is designed for mixed-experience cohorts: it starts with clear conceptual foundations around LLMs, context windows, prompting boundaries, and harness-loop style agents, then moves into practical Codex workflows with tasks covering common commands, built-in commands, custom commands, skills, subagents, and memory, and finishes with advanced engineering patterns such as MCP integrations, BMAD-style team workflows, and spec-driven development inspired by GitHub Spec Kit. By the end, participants will be able to use Codex as a serious engineering tool rather than a chat novelty, structure work for reliable execution, and design agentic delivery flows that scale from solo experimentation to disciplined team practice.

Agenda

Workshop Agenda

Day 1: Foundations for Agentic Coding

What LLMs are, how they reason, and where they fail
Tokens, context windows, retrieval boundaries, and prompt compression
From autocomplete to agents: loops, planning, tools, and state
The harness loop concept: perception, decision, action, verification
When to stay manual and when to delegate to an agent
First guided exercises without full task automation

Day 2: OpenAI Codex Fundamentals in Practice

Codex mental model: goals, context, edits, and execution
Common commands for navigation, inspection, and implementation
Built-in commands and safe ways to use them in day-to-day work
Custom commands, repeatable workflows, and reusable operator habits
Skills, subagents, and memory as force multipliers
Hands-on tasks that move from small edits to multi-step implementation

Day 3: Advanced Agentic Engineering Workflows

MCP architecture: tools, contracts, schemas, and trust boundaries
Designing tool-using agents that stay observable and safe
BMAD-style collaboration and role-based AI workflows
Spec-driven development with a GitHub Spec Kit mindset
Turning requirements into plans, plans into execution, execution into review
Capstone: deliver a spec-first Codex workflow with tools, memory, and verification

Starting from

€1,390.00

Topics

Agentic Software Engineering

Customizable to your team's needs

Expert trainers with industry experience

Hands-on exercises and real-world projects