Melbourne AWS User Group #165 - November 2026
Details
🎉 Melbourne AWS User Group – November Meetup #165 🎉
📅 Wednesday, 25 November
📍 NAB Events Hub, Ground Floor, 395 Bourke Street, Melbourne
⏰ Food at 6:00 pm, talks from 6:30 pm AEDT
We’re back to our regular format for our final meetup of 2026! Join us for What’s New in AWS and two talks on putting generative AI to work.
Aidan Grace will share what he’s learned deploying AI in organisations where security comes first. Meiyappan Chidambaram will run one question three ways to show how hybrid search and semantic caching work together in a RAG system on AWS.
A huge thank you to our sponsors 🥇 Mantel Group (Gold), 🥈 Cevo (Silver) and 🏛 NAB (Venue).
Can’t make it in person? Watch the livestream on YouTube.
***
🗓️ Agenda
- 6:00 pm – Food & Socialising
- 6:30 pm – Introductions
- 6:40 pm – What’s New in AWS
- 6:55 pm – The Kingdom That Switched On the Magic (Level 200) with Aidan Grace
- 7:25 pm – Advanced RAG at Scale (Level 300) with Meiyappan Chidambaram
- 7:55 pm – Marketplace (Who’s Hiring)
- 8:00 pm – Networking + Close
***
🎤 Talk Details
The Kingdom That Switched On the Magic: Lessons from Deploying AI Where Security Comes First
Aidan Grace, Head of Applied AI at Cloudscaler · Level 200
The kingdom of Orgalot wants AI switched on by Tuesday. A patient Tortoise has a different timetable, starting with governance, then enablement, then getting AI into production.
From there, Aidan shares five lessons from deploying generative AI with Amazon Bedrock at the UK Health Security Agency and other UK organisations. He’ll cover why the data swamp can wait, what to measure instead of headcount savings, and how to build guardrails people will actually use. He’ll finish with a look at how Cloudscaler tracks AI usage, value and cost to answer the King’s question. Is the Magic working?
***
Advanced RAG at Scale: Hybrid Search and Semantic Caching on AWS
Meiyappan Chidambaram, Senior Cloud Consultant at Cevo Australia · Level 300
Watch one question run three ways through a RAG system built on Amazon Bedrock, OpenSearch Serverless and MemoryDB.
- A new question misses the cache and runs the full pipeline, from query rewriting and hybrid BM25 and vector retrieval through rank fusion and reranking to a cited answer.
- A paraphrase hits the semantic cache and returns in milliseconds without a model call.
- The same question with a different metadata filter deliberately misses the cache, showing how filters control when an answer can be reused.
***
🙌 Want to speak at a future meetup?
We’re always looking for new speakers. If there’s something you’d like to share, come have a chat on the night or get in touch through our website.
See you at NAB for one last meetup before the break!
