Guardrails

Goal: Learn to build input and output guardrails, PII redaction, content moderation, and prompt-injection defenses, the practical controls that keep an LLM application safe once real users and real data show up.

Do:

1. Watch the video above ⬆️

2. Build your own guardrails hands-on https://github.com/NVIDIA-NeMo/Guardrails

Bonus read: OWASP Prompt Injection Prevention Cheat Sheet (covers guardrail models like Llama Guard and Granite Guardian) https://cheatsheetseries.owasp.org/cheatsheets/LLM_Prompt_Injection_Prevention_Cheat_Sheet.html