In this training you learn to develop applications with language models that access your own documents and give dependable answers, and it is aimed at developers with Python experience. You select suitable context, connect sources through retrieval-augmented generation, use structured output and check answer quality with test cases. Cost, latency and security risks such as prompt injection are covered as well. Platform and scope are agreed with you when you enquire.
What you’ll take away.
- Describe the building blocks of a language model application: prompt, context, model and post-processing
- Split, embed and store documents in a vector store for retrieval
- Select relevant context and generate answers with source references
- Use structured output and tool calls with schema validation
- Measure answer quality with a test set, metrics and manual sampling and detect regressions
Your training in detail
- Who is it for?
- Software developers, architects and technical product owners building document assistants, search features or automation with language models.
- Prerequisites
- Python skills, experience with REST APIs and JSON, and some first contact with a language model.
- Format & planning
- Suggested format: 3 days live online or on site, split into modules. Scope depends on prior knowledge.
Agenda: from understanding to application
Our independently developed agenda proposal. Focus areas, lab environment and scope are agreed before booking.
Architecture of a language model application: prompt templates, system instructions, context windows and model choice.
Document preparation with parsing, chunking, metadata and embeddings for the vector store.
Retrieval-augmented generation: retrieval, reranking, context selection and answers with citations, for example with LangChain.
Structured output with JSON schema, function calling and validation with retry on errors.
Quality checks with test questions, reference answers, faithfulness to context and review by human and model.
Operations and security: cost, latency, caching, logging and protection against prompt injection per OWASP.
Suggested exercise: You build a question-answering assistant over a set of manuals, have it answer with citations and test it with 25 questions you write yourself before adjusting chunk size and retrieval count.
Exams, licences & credentials
There is no single authoritative external exam for these topics. Attending alone awards no vendor certificate. Exam fees and licences are only included if the offer confirms them.
Vendor references & review date
Sources checked on 2026-11-03. These sources describe the technology or external exam requirements. They do not imply a partnership with NextNowa.
Your learning content
These objectives describe the training topic. The specific syllabus, duration, dates, trainer and price are agreed before booking.
Frequently asked questions
What is RAG and when do I need it?
RAG stands for retrieval-augmented generation: before answering, the application looks up relevant passages in your documents and passes them to the model as context. It can then use current, proprietary knowledge and cite sources. It makes sense for questions about internal documents the model does not know.
RAG or fine-tuning: which is better for proprietary knowledge?
For factual knowledge that changes, RAG is usually the better choice: you update documents rather than the model. Fine-tuning suits style, format or fixed tasks better. The two are often combined only later. Start with RAG and good prompting and measure whether that is enough for your case.
How do I check whether my application's answers are good?
You build a test set of real questions with expected answers or sources and measure after every change: was the right passage found, does the answer stay with the context, is it complete? You also review samples by hand. That way you catch regressions early.
Does Building language model applications include a certificate?
There is no single authoritative external exam for these topics. Attending alone awards no vendor certificate. Exam fees and licences are only included if the offer confirms them.