Call for submissions: Platform Engineering meet-ups
Rootconf
Rootconf

Call for submissions: Platform Engineering meet-ups

Share your journey of building platforms that power engineering teams

All submissions
Abhijit Roy

Abhijit Roy

@Roy214

Guardrails: Keeping a General-Purpose Model in Its Lane

Submitted Aug 19, 2026

Session type: Talk (30 mins) Submission made for: Bangalore meet-ups

{Describe your session in 2 paragraphs}

You deploy an LLM as a history assistant and it cheerfully writes Python. You ship a support bot and someone talks it into selling a Tahoe for a dollar. General-purpose models will answer anything, leak anything, and follow whatever the last person in the conversation told them to do. This session covers the layer that sits between your users and your model and decides what actually gets through — prompt injection, system prompt leakage, PII exposure, hallucinated facts, and off-topic drift — and why a better system prompt stops working the moment someone is actually trying.

Then we get hands-on with NVIDIA NeMo Guardrails on OpenDaathub/Red Hat OpenShift AI: input and output rails, pre-built detectors (regex, Presidio PII, jailbreak heuristics, HF classifiers), LLM-as-judge for custom policy, the two API surfaces, deployment via the TrustyAI operator, and a live demo. We close on monitoring with OpenTelemetry — how much latency you’ve added, which detector is the bottleneck, and why a request got blocked.

{Mention 1-2 takeaways from your session}

  • A system prompt is not a security boundary. Attendees will leave knowing which failure modes actually need a separate enforcement layer — prompt injection, PII leakage, scope drift — and how to structure input and output rails so the check runs outside the model rather than inside its instructions.

  • You can build this on open source, inside your own perimeter. A working mental model for deploying NeMo Guardrails on OpenShift AI: which pre-built detector to reach for versus when to write an LLM-as-judge policy, and how to use OpenTelemetry traces to measure the latency cost and debug why a request got blocked.

{Which audiences is your session going to beneficial for?}

Platform / MLOps engineers, AI engineers

{Add your bio - who you are; where you work}

I am a SSME at Red Hat. My work centers on making large language models (LLMs) more efficient, scalable, and production-ready.

All submissions

Comments

{{ gettext('Login to leave a comment') }}

{{ gettext('Post a comment…') }}
{{ gettext('New comment') }}
{{ formTitle }}

{{ errorMsg }}

{{ gettext('No comments posted yet') }}

Register

Hosted by

Rootconf
Rootconf
We care about site reliability, cloud costs, security and data privacy