Building LLM-Powered Applications
Architecture, evaluation and guardrails for production
By Houssam Kodad
One-time purchase
€31.95
VAT included
where applicable
- Instant download after purchase
- Readable on any device
- Free updates to this edition
- Secure checkout
About this book
What's inside
A working demo with a language model takes an afternoon; a product people rely on takes engineering. This book covers that engineering: structuring prompts and context, orchestrating tool calls, measuring whether the system is actually correct, and putting guardrails in place before users find the edge cases. It's model-agnostic and grounded in patterns that outlast any single provider's API.
What you'll learn
Skills you'll walk away with
- Design application architecture around an LLM
- Engineer prompts and manage the context window
- Get reliable structured outputs and function calls
- Orchestrate tools and multi-step agent flows
- Build evaluation suites that measure real quality
- Add guardrails for safety, privacy and failure modes
- Control latency and cost under real traffic
- Observe, debug and continuously improve the system
Table of contents
10 chapters-
01
From Demo to Dependable Product
- · What changes at production scale
- · A reference architecture
- · Choosing and abstracting models
-
02
How LLMs Behave and Fail
- · Tokens, context and limits
- · Hallucination and its causes
- · Determinism and temperature
-
03
Prompt and Context Design
- · System, user and tool messages
- · Few-shot and templating
- · Context-window budgeting
-
04
Structured Outputs and Function Calling
- · JSON schemas and validation
- · Tool/function definitions
- · Handling malformed responses
-
05
Tool Use and Agents
- · Single-step tools
- · Multi-step planning loops
- · Stopping conditions and loops
-
06
Evaluation You Can Trust
- · Building a labelled eval set
- · LLM-as-judge and its pitfalls
- · Regression testing prompts
-
07
Guardrails and Safety
- · Input and output filtering
- · PII and data handling
- · Jailbreaks and prompt injection
-
08
Latency, Caching and Cost
- · Streaming and partial results
- · Prompt and semantic caching
- · Model routing and fallbacks
-
09
Observability and Improvement
- · Tracing requests end to end
- · Capturing feedback
- · Closing the improvement loop
-
10
Shipping and Operating
- · Rollouts and versioning prompts
- · Incident response
- · A production checklist
This is the full chapter list — exactly what you'll receive in the PDF.
More in AI & LLMs
Keep exploring this track
Retrieval-Augmented Generation in Production
Chunking, embeddings, vector search and reranking that actually work
Fine-Tuning and Adapting Open LLMs
LoRA, quantization and instruction tuning on your own data
Prompt Engineering for Developers
Reliable patterns, structured outputs and tool use