Course Objective & Scope
Relying on a single system prompt rule or content filter cannot protect enterprise LLM applications; effective defence requires complementary deterministic and model-based controls across input, context/retrieval, tool execution and output stages. This course covers building multi-layer guardrail architectures and runtime policy enforcement using Llama Guard, Azure AI Content Safety, NeMo Guardrails and Guardrails AI.
Hands-on sessions address synchronous and streaming inspection patterns, latency budget optimisation and false-positive/false-negative trade-offs, culminating in a guardrail and policy enforcement plan tailored to participants' systems.
Who Should Attend?
- AI and backend engineers deploying LLM and agent applications to production
- Teams building enterprise AI gateways and platform infrastructure
- Application security engineers and security architects
- Prerequisite: Python service development and prior LLM API integration experience
Key Takeaways
- Design defense-in-depth layers across input, retrieval, tool calling and output stages
- Combine deterministic rule engines with lightweight classifiers and judge LLMs
- Write topic control and schema validation policies with NeMo Guardrails or Guardrails AI
- Implement parallel guardrail inspection for streaming responses without blowing latency budgets
- Measure false-positive, false-negative and p95 latency metrics across guardrail layers
Curriculum
Defense-in-Depth and Guardrail Architecture
- Control points across input, context, tool invocation and output stages
- Dividing work between deterministic rules, BERT/DeBERTa classifiers and LLM-as-a-judge layers
Input and Topic Control Layers
- Detecting prompt injection, toxic content and off-policy/competitor topics
- Constraining dialogue rails with NeMo Guardrails (Colang) and Llama Guard
Output Validation and Groundedness / Policy Checks
- Structured output, format and business rule validation with Guardrails AI
- Source groundedness verification and PII/secret leakage filters
Tool Calling and Action Policy Enforcement
- Runtime parameter schema, business rule and quota checks
- Safe fallback responses, circuit breakers and escalation workflows
Performance, Streaming and Operational Monitoring
- Chunk-based asynchronous inspection and latency optimisation for streaming responses
- Guardrail telemetry, bypass analysis and continuous rule tuning
Other Courses in This Category
-
Teaches how to build automated security evaluation suites for LLM and agent applications using promptfoo, garak and Inspect AI, enforce regression gates in CI/CD pipelines and continuously test model and prompt changes.
- Live Course
- 5 modules
-
Teaches how to design enterprise LLM, RAG and agent platforms on Zero Trust principles using a central AI gateway, network and tenant isolation, KMS key management, secure MCP integration and auditable telemetry.
- Live Course
- 5 modules
-
Comprehensive training on PII masking, reversible pseudonymisation, anonymisation and data residency architectures to adopt LLMs without exposing personal or confidential corporate data.
- Live Course
- 5 modules
-
Hands-on training on building a low-latency PII detection, masking and de-pseudonymisation gateway in front of LLM APIs using Microsoft Presidio, GLiNER and Turkish rule/NER recognisers.
- Live Course
- 5 modules
-
Teaches how to replace production customer data in testing, analytics and LLM evaluation environments using k-anonymity, differential privacy and SDV/LLM-based synthetic data generation.
- Live Course
- 5 modules
-
Teaches how to scrub personal data from LLM and agent telemetry in OpenTelemetry, Langfuse and SIEM logs, and engineer retention policies, access controls and KVKK compliance into observability pipelines.
- Live Course
- 5 modules
Get in touch
