RSVP on our GDG community platform -> https://gdg.community.dev/events/details/google-gdg-melbourne-presents-gdg-melbourne-june26-meetup/

This event is hosted at the Mantel Group Office, with the recording of the talks to be shared afterwards on YouTube if you can't make it.

There will be food, networking and some surprise swag!

Agenda:

6:20 - Intro

6:30 - Talks

8:00 till late - networking & chat (& drinks optional)

Speakers and Talks:

Talk 1: Olga Mirensky - Senior Staff Site Reliability Engineer @ Ping Identity

Olga and Her Agent Attended AI Engineer Conference. Lessons Learned

What happens when you let an experimental Stand-In Agent run loose against a live conference environment? To follow the fast-paced sessions of an AI Engineer conference, I built an agent while at the conference. The agent monitored talks and distilled what actually mattered to me - filtering for high-value signal and turning it into specs ready to implement or articles I should read.
I started with a different platform, but pivoted to an ADK agent deployed to GCP Agent Platform. Identical system prompt, different platform architecture, different LLM - completely different outcomes.
This talk is a transparent look at what happens when agentic experiments meet reality, including an architectural runaway bug that burned 1M tokens per minute.
We look past the hype and focus on the real engineering rigour required to build purposeful agents, with four key takeaways:

* The Purpose Filter: Why does your agent care about this specific talk? If it’s not highly unique, you’re just automating noise.
* Architecture > Hype: The gap between what you expect and what actually happens at runtime and why.
* The Personalised Flywheel: High-signal distillation requires deep context about you, your team, and where you are in your journey.
* Clear Goal Setting: Moving from “autonomous” to “purposeful”.

In the replay demo I’ll walk through my flywheel showing how the agent filtered the content in real-time, picked up on tech I am interested in, turned it into an implementable spec, and how AI tools can close the loop to further automate the entire implementation, testing, and release cycle.

Talk 2: Sahil Bahadur Devkota - Full Stack Developer

Beyond LLMs: why RAG is equally important

In this talk, I'll break down how RAG works, why is it really important in real-world and it's popular architectural pattern. We'll look at the core components of a RAG pipeline, discuss it's implementation approach, and cover some of the key things to think about when building an AI applications with RAG.

Whether you're working with AI or building application that need accurate responses, you'll leave with a practical understanding of when RAG make sense - and when it doesn't.

Talk 3: Abhijeet Kumar - AI & Automation | Cloud enthusiast

Beyond RAG: Building an Agentic "Second Brain" with FastAPI, pgvector, and Gemini

Move beyond static note-taking apps. This talk explores Project NeuralFortress—a custom, Python-built knowledge management system designed to actively synthesize information. We’ll walk through evolving a basic RAG script into an agentic pipeline using FastAPI for routing, Neon PostgreSQL (pgvector) for vector storage, and Google Gemini for cognitive processing.

In Development (Sneak Peek):
A brief look at experimental features currently in testing to make the system fully autonomous:
GraphRAG: Mapping multi-hop relationships beyond standard vector distance.
MCP Integration: Native database queries directly from your code editor.
Async Watchers: Autonomous background web scraping and data ingestion.

How to find us:

Doors open at 5:45 PM at our host the Mantel Group, Level 2/452 Flinders St, Melbourne. Message via Meetup or Slack if you are arriving after 6:00, as the lifts won't let you up. 

This is an in-person only event. If you have any accessibility requirements, please reach out via the Meetup or Slack to the organiser group to discuss, and we will do our best to help you out.