Building production LLMs is like herding cats. 🐱🤖
In this talk, Diane Loi shares a practical guide to AI guardrails: how to enforce safety, evaluate trade-offs, and test systems designed to surprise you.
Watch: youtube.com/watch?v=qYcZ5e…
#AI #LLM #SoftwareEngineering
Programmable - the Australian software engineering conference
Joined July 2022
- 1 house, 4 different database entries? 🏠 Why do data sources never agree on addresses?In this Programmable 2026 talk, Olga Shabalina breaks down Entity Resolution & data modeling to show how data engineers untangle real-world data messes. Watch: youtube.com/watch?v=ie140j…
- Securing LLMs was about what they say. With Agentic AI, it's about what they do. 🛡️ Toby Allen’s 4-pillar security blueprint for autonomous agents: 1️⃣ Non-human auth 2️⃣ Delegated permissions 3️⃣ Human-in-the-loop 4️⃣ Scoped RAG Watch: youtube.com/watch?v=IDW8PC… #AgenticAI
- The bottleneck isn't writing code. It's waiting to safely test it. Dan Baker explains how to fix that at scale — without breaking shared envs. youtu.be/hORYOBDvNXs #PlatformEngineering #Programmable2026
- Send a selfie on WhatsApp. Get back a cartoon. That's it. That's the talk. 🎨 @twilio + OpenAI image APIs = creative AI experiences in a chat. Watch → youtu.be/Xh3d0aFb1G0 #Programmable2026

