Question 1

What AI for Developers and Engineers training courses does JBI Training offer?

Accepted Answer

JBI Training offers ten specialist AI courses designed specifically for software developers and engineers who need to build, integrate, test, secure, and operate AI systems professionally. Available courses are Build AI Agents with Python, RAG and MCP (two days), RAG: AI on Your Own Data (two days), AI Integration in Legacy Codebases (two days), Testing and Evaluating AI Outputs (two days), AI Security: Attacks and Defences (two days), Self-Hosted Models for Sensitive Environments (two days), Building AI Agents with Real APIs (two days), Agentic Coding with Claude Code (three days), Agentic Coding with GitHub Copilot (three days), and AI-Assisted Python (one day). All courses are available as scheduled classroom sessions in London, as live online instructor-led training, or as customised onsite programmes for software teams.

Question 2

What is RAG and what does the RAG: AI on Your Own Data course cover?

Accepted Answer

RAG stands for Retrieval-Augmented Generation — a technique for making AI language models answer questions using your own documents, databases, or knowledge bases rather than relying solely on the general knowledge they were trained on. It works by retrieving the most relevant chunks of content from a private data store at query time and supplying them as context to the language model alongside the user's question, enabling accurate, grounded responses from proprietary data. JBI's two-day RAG: AI on Your Own Data course covers the architecture and components of RAG systems, chunking and embedding strategies, vector databases, retrieval pipeline design, prompt construction for RAG, evaluation of retrieval quality, and building a working RAG application. It is suited to developers and engineers who need to build AI systems that work reliably with internal or sensitive data.

Question 3

What is MCP and why is it relevant to building AI agents?

Accepted Answer

MCP (Model Context Protocol) is an open protocol developed by Anthropic that standardises how AI models connect to external tools, data sources, and services. It defines a consistent interface through which AI agents can call tools — such as web search, database queries, file systems, or custom APIs — making it easier to build composable, interoperable AI systems without writing custom integration code for every tool. JBI's Build AI Agents with Python, RAG and MCP course covers how to use MCP alongside Python and RAG techniques to build AI agents that can reason over private data and take actions across real systems. Understanding MCP is increasingly important for developers building production-grade AI agents as the protocol gains adoption across the AI tooling ecosystem.

Question 4

What does the AI Integration in Legacy Codebases course cover?

Accepted Answer

The AI Integration in Legacy Codebases course is a two-day programme for developers who need to add AI capabilities to existing software systems — rather than building new AI-native applications from scratch. It covers the challenges specific to integrating AI into legacy or established codebases, including how to identify appropriate integration points, how to wrap AI API calls safely within existing service architectures, managing latency and reliability when calling external AI services, testing and monitoring AI-enhanced functionality in production, handling data privacy and compliance considerations when passing existing data to AI models, and managing the ongoing cost and performance of AI API usage at scale. It is suited to senior developers and architects working on existing enterprise systems.

Question 5

What does the Testing and Evaluating AI Outputs course cover?

Accepted Answer

Testing AI systems requires different approaches from testing conventional software — AI outputs are probabilistic rather than deterministic, making traditional unit testing insufficient on its own. JBI's two-day Testing and Evaluating AI Outputs course covers how to design and implement evaluation frameworks (evals) for AI systems, how to define meaningful metrics for different AI task types — including accuracy, relevance, groundedness, and safety — how to build automated evaluation pipelines, how to detect regression in AI output quality over time, human evaluation approaches for subjective tasks, and how to use evaluation results to guide prompt engineering and model selection decisions. It is suited to developers, QA engineers, and MLOps professionals responsible for the reliability and quality of AI systems in production.

Question 6

What does the AI Security: Attacks and Defences course cover?

Accepted Answer

The AI Security: Attacks and Defences course is a two-day programme covering the security vulnerabilities specific to AI systems and how to defend against them. It covers prompt injection attacks — where malicious input manipulates an AI agent's behaviour — jailbreaking techniques, data poisoning, model extraction, adversarial examples, indirect prompt injection through external content retrieved by RAG systems, and supply chain risks in AI pipelines. The course also covers defensive techniques including input validation and sanitisation, output filtering, privilege separation in agentic systems, monitoring for anomalous AI behaviour, and responsible disclosure and governance frameworks. It is suited to developers building AI-integrated applications and security engineers responsible for AI systems in production environments.

Question 7

What does the Self-Hosted Models for Sensitive Environments course cover?

Accepted Answer

The Self-Hosted Models for Sensitive Environments course is a two-day programme for developers and engineers who need to run AI language models within their own infrastructure — rather than sending data to external cloud AI APIs — due to data sensitivity, regulatory compliance, air-gapped environments, or cost considerations. It covers the landscape of open-weight models suitable for self-hosting, hardware and infrastructure requirements for running models locally or on private cloud infrastructure, tools and frameworks for model serving including Ollama, vLLM, and llama.cpp, integrating self-hosted models into existing application architectures, performance and latency considerations, and governance and access control for private model deployments. It is particularly relevant for organisations in regulated industries such as finance, healthcare, defence, and the public sector.

Question 8

What is the difference between the Agentic Coding with Claude Code and Agentic Coding with GitHub Copilot courses?

Accepted Answer

Both are three-day courses covering professional agentic coding workflows, but each focuses on a different tool. Agentic Coding with Claude Code covers how to work professionally with Anthropic's Claude Code — a terminal-based agentic coding tool that can read, write, execute, and test code autonomously within a project. Agentic Coding with GitHub Copilot covers GitHub Copilot's agent mode and workspace features, which integrate directly into VS Code and GitHub workflows. Both courses cover the same core professional disciplines — spec-driven development, test-driven verification of AI-generated code, steering and evaluating agent behaviour, and integrating agentic tools into a professional development workflow — applied to their respective tool. The right course depends on which tool the developer or team uses or plans to adopt.

Question 9

What is Building AI Agents with Real APIs and who should attend?

Accepted Answer

The Building AI Agents with Real APIs course is a two-day programme that teaches developers how to build AI agents that interact with real, production APIs — rather than toy examples or sandboxed environments. It covers how to design agent architectures that call external APIs reliably and safely, how to handle authentication, rate limiting, error handling, and retries in agentic API workflows, how to structure tool definitions for AI agents, how to test and validate agent behaviour across real API responses, and how to build observability into agents that make API calls in production. It is suited to backend developers and software engineers who are building AI-powered features or services that need to interact with third-party systems, internal APIs, or enterprise service layers.

Question 10

Can these AI for Developers courses be customised for a corporate software team?

Accepted Answer

Yes. All courses in this group can be delivered as customised closed-group programmes for corporate software and engineering teams, onsite at your organisation's premises or online. Content and lab exercises can be tailored to the team's existing technology stack, the AI tools and APIs they are working with, their codebase language and architecture, and the specific AI integration challenges they are working to solve. Organisations building internal AI products, integrating AI into enterprise systems, or establishing professional AI development standards across a team will find the customised delivery format particularly effective. JBI has delivered AI and software engineering training for development teams at organisations including the BBC, NHS, RBS, Sky, EDF, and Cisco.

Question 11

Are these courses available online?

Accepted Answer

Yes. All JBI Training courses run as live online instructor-led sessions as well as in-person in London and as bespoke on-site training at your organisation.

Question 12

Can these courses be customised for my team?

Accepted Answer

Yes. All JBI Training courses can be delivered as bespoke closed-group programmes tailored to your team's technology stack, experience level, and business objectives. Contact enquiries@jbinternational.co.uk.

Developers and Engineers

Practical, lab-based courses for software teams working with AI APIs, existing codebases, and enterprise systems.

Build AI Agents with Python RAG and MCP

RAG: AI on Your Own Data

AI Integration in Legacy Codebases

Testing and Evaluating AI Outputs

AI Security: Attacks and Defences

Self-Hosted Models for Sensitive Environments

Building AI Agents with Real APIs

Agentic Coding with Claude Code

Agentic Coding with GitHub Copilot

AI-Assisted Python

FAQs