Thoughts During Tuition Free Week

A short diary entry during Curtin’s tuition free week.

From 3 April, Curtin had a two-week tuition free week, which gave me some time to pause and catch up on study. It felt like a much-needed break in the middle of the semester, but also a good chance to use the time well instead of letting it slip away.

One thing I was especially glad about was being able to attend the free Agentic AI Conference run by Data Science Dojo. Since it was scheduled in Pacific Time, I could not watch most of it live, but I followed along later through the recorded YouTube sessions. Even though it was not real-time, it still felt valuable because the content itself was so relevant to what I have been thinking about lately.

I was first introduced to agentic AI during my internship, so it was exciting to revisit the topic through a free workshop and actually build a distributed multi-agent system myself. What I found most interesting was that the system was not based on a single LLM doing everything. Instead, the work was divided across multiple specialised agents, each with its own role.

The system we built included four main agents:

• Researcher — gathers up-to-date information through web search
• Judge — critiques and checks the quality of the research
• Content Builder — turns the results into a structured course
• Orchestrator — manages the overall workflow

During the workshop, we experimented with several useful ideas, including web-search-enabled agents, structured outputs with Pydantic, remote agent connections through an agent-to-agent protocol, iterative feedback loops between the Researcher and Judge, running the distributed system locally with ADK, and finally deploying it to Google Cloud Run.

The overall design used a combination of SequentialAgent and LoopAgent, where the Researcher and Judge continued to interact until the output reached an acceptable standard, and then the Content Builder generated the final result. It was a good reminder that agentic systems are not just about having a stronger model, but also about designing the workflow around reasoning, validation, and collaboration.

During the same break, I also volunteered at the World Quantum Day event run at Pawsey Supercomputing Centre. I could only help in the morning because I had an Honours research Teams meeting in the afternoon, but even that short time was really memorable. It was nice to meet PhD students and researchers, and I especially enjoyed seeing how engaged the children were with the activities.

I was helping at the quantum gates corner, where the activity was set up like a maze game. The children had to use gates such as NOT and Hadamard correctly in order to match the expressions and open the maze doors. At first, I thought it might be too difficult for younger kids, but they surprised me. Watching them solve each level one by one, with a bit of guidance, was genuinely rewarding. I had a lot of fun explaining the ideas and seeing their excitement when they got the answers right. Time passed much faster than I expected.

Aside from those events, I have also been working on HPC Assignment 2, while continuing to look for opportunities such as graduate roles starting in 2027. There is always a lot going on at once, but in a way I am grateful for that. My Honours research is interesting, and I am also really enjoying my High Performance Computing elective this year. Both make me feel that I am moving in the right direction.

Overall, this tuition free week gave me more than just time off. It gave me space to reflect, to learn outside the classroom, and to reconnect with the reasons I enjoy this field. There is still a lot I want to improve, and a lot I am uncertain about, but I want to keep going steadily and finish this year well.