Trainings

At Playful Python, we deliver hands-on, high-engagement training programs in generative AI and developer productivity.

Our workshops are uniquely designed with at least 50% of time dedicated to lab activities and interactive discussions. Participants build real, working projects that they can take back and apply to their own work.

Our courses are structured to help engineers of all levels gain confidence and practical expertise in modern development tools and techniques.

Additionally, we are open to tailor our course material to the particular needs of your organisation, ensuring the best experience for your engineers.

Learning Paths

At Playful Python, we offer various modules that can be combined into customised learning paths. This allows organisations to easily select the specific topics that are relevant for them.

Building AI Applications Learning Path

This following modules are available for engineers who are building AI applications:

• AI-Enhanced Coding (2 day)
• Prompt Engineering for AI developers (1 day)
• Introduction to Retrieval Augmented Generation (1 day)
• Retrieval Augmented Generation Deep Dive (2 day)
• Introduction to Agentic Applications (1 day)
• Agentic Applications Deep Dive (2 day)

These modules can be taken individually, or combined together as required into a customised learning path.

AI-Enhanced Coding

This two-day course introduces developers to cutting-edge AI coding assistants such as GitHub Copilot, Claude Code and Cursor.

Participants will learn how to integrate these tools into their development workflows to accelerate feature design, implementation, and documentation.

Coding assistants have a reputation for generating poor quality, unmaintainable code which leads to a sharp increase in maintenance costs a few months later. In this course we will put special focus on best practices so that AI assisted code sits alongside human generated code at the same level of quality.

Learning Outcomes

• Learn how to understand new or unfamiliar code bases using AI tools
• Learn to use specialised IDEs (Cursor), Extensions (Github Copilot) and terminal AI agents (Claude Code)
• Accelerate feature development using agentic coding agents
• Understand best practices so that AI generated code is clean, readable and adheres to your coding standards
• Learn AI tooling across the development lifecycle, including design, implementation, testing, code review, documentation and bug fixes

Course Outline

• Installation of tools and basic tool overview
• Chat mode versus Agentic mode
• Use chat mode to understand unfamiliar codebases
• Generate design and architecture documents for existing projects
• Collaborate with Copilot to plan and implement new features
• Utilise the agentic mode to automate coding tasks, perform testcase generation, and update documentation
• Learn how to connect AI tools to external services using Model Context Protocol (MCP)
• Use MCP to review pull requests on Github and update tickets on Jira
• Build subagent workflows for multi-step agentic coding tasks
• Best practices to minimize hallucinations and coding errors

Prompt Engineering for AI Development

Prompt Engineering is the cornerstone of AI development. Effective prompting strategies can be the difference between reliable and unreliable AI applications.

In this class, we will learn prompt engineering techniques that make a difference to your application.

Learning Outcomes

• Learn how to get control of LLM behaviour
• Understand prompt engineering techniques
• Learn how to incorporate thinking and reasoning
• Understand differences between regular LLMs and reasoning models

Course Outline

• Understand how LLMs work under the hood
• Learn LLM control parameters: temperature, top_p, top_k
• Know the different types of models: Base models, instruct models and reasoning models
• Basics of prompt engineering
• Incorporating reasoning into the prompt
• Use structured outputs and tool functionality

Introduction to Retrieval Augmented Generation

One of the basic limitations of LLMs are that they cannot answer on topics that they have not been trained on. Therefore LLMs cannot answer questions based on the large corpus of internal documents that organisations possess. The way around this is retrieval augmented generation or RAG.

In this class we explore the basics of RAG, how it works, and how to build a RAG application.

Learning Outcomes

• Learn the limitations of LLMs in processing internal corporate datasets
• Understand the components of the RAG architecure
• Build a RAG application from scratch – no frameworks used
• Deploy a RAG application to production

Course Outline

• What are the limitations of LLMs
• How do we overcome these limitations with RAG
• Components of a RAG application
• Understanding embeddings and vector databases
• Ingestion and chunking
• Retrieval and generation
• Deploying RAG applications

Retrieval Augmented Generation Deep Dive

The RAG deep dive course builds on the introductory RAG course, to delve into more advanced topics that arise when trying to build production grade applications.

In this class we explore the various failure modes of a RAG application, how to effectively evaluate the application to determine the points of failure, and techniques to fix the points of failure.

Learning Outcomes

• Build production ready RAG applications
• Learn how to evaluate a RAG application
• Understand all failure modes for the application
• Learn techniques for overcoming the failure modes

Course Outline

• Failure modes in RAG applications
• Evaluating a RAG application
• Advanced chunking strategies
• Handling multimodal content
• Dealing with Docs, PDFs and PPTs
• Query understanding and rewriting
• Intent routing
• Improving retrieval quality
• Hybrid search
• Graph RAG
• Agentic RAG
• Balancing cost, latency and accuracy

Introduction to Agentic Applications

The promise of AI is to build applications that can sense, reason, plan and take decisions. In this introductory class on building Agentic applications, we explore exactly that. We'll learn how agentic applications work under the hood, and how to create agents using the Langgraph framework.

Learning Outcomes

• Learn how agents work from first principles
• Get an understanding of the Langgraph framework
• Learn to build single and linear multi agent applications

Course Outline

• Understand tool calling and structured outputs
• Framework of agents as LLMs in a loop
• Introduction to Langgraph
• Building an agent in Langgraph
• Linear multi-agent systems
• Deployment to production

Agentic Applications Deep Dive

This class follows on from the introductory class to do a deep dive into the components that power agentic applications. We will learn how to evaluate agents, enhance agents with long and short term memory and integrate agents with MCP server. We will also learn about orchestrating multi-agent systems.

Learning Outcomes

• Find out how to improve reliability by evaluating agents and pinpoint failures
• Increase agent safety by implementing human-in-the-loop architecture
• Implement multi agent systems with reasoning and planning

Course Outline

• Workflows and Agents
• Parallel, sequential and hierarchical task orchestration
• Multimodal agents
• Memory: Short term, long term and entity memory
• Testing and evaluating agentic systems
• Evaluation metrics: Task adherence, tool call accuracy
• Synthetic testset generation
• Human-in-the-loop architectures
• Monitoring and observability
• Integrating with tracing systems
• Scaling and performance