“I’m in love with the idea of this project.”
— Santiago Massé, Technical Coordinator of the HIVEMIND project
Santiago Massé, Technical Coordinator of HIVEMIND and researcher at idener, speaks with genuine enthusiasm about the project’s mission and future. In this interview, he discusses how his expertise in language technologies supports the project’s core objectives, why he believes in its long-term value, and how the team is building a flexible, future-oriented framework for developers in the age of generative AI.
Watch the interview
Prefer reading?
You can find a full transcript of the interview below.
I’m Santiago Massé, a computer scientist, and now also a project coordinator. We’re coordinating this project, but we’re also developers in several work packages, specifically Work Package 4 and Work Package 7. We’re currently setting up the requirements gathering phase of the project to ensure the execution actually works. That’s our main role.
About myself, my main interest is in language technologies, and I’ve been researching this area in recent years. I’m really happy that this project came together.
How does your expertise contribute to the objectives of HIVEMIND?
My expertise contributes mostly in Work Package 4, which is the most closely related to fine-tuning large language models and ensuring they behave as we expect. This is something we’re really striving for, training, fine-tuning, and using LLMs in a way that fits human needs.
What excites you most about being part of the project?
Honestly, I wrote most of the proposal, and I never thought it would make it this far. But from the beginning, when I was presenting it to the rest of the partners and inviting them to join the consortium, I told them, I’m in love with the idea of the project.
It might seem super ambitious, I think it is, but it’s so futuristic that I believe it can work. And I believe it will work.
What impact do you expect HIVEMIND to have, and what challenges lie ahead?
It’s a project related to generative AI, which is evolving at an incredible speed. From the start, we were aware that we needed to create a framework that could adapt to this fast-changing landscape.
I think the impact on software developers will be significant, because the project is built to evolve, by integrating open-source LLMs as they emerge. Anyone working in the LLM space knows how quickly things change. One year ago, we were talking about LLaMA 2, now there are five different open-source alternatives that seem to outperform it.
So yes, I think this project will benefit software developers by providing a safer, more structured way of applying generative AI to software engineering, something that’s currently missing in the available tools and approaches.