To build the best tech products, we need to stay up-to-date on the latest research and trends. Actually, keeping up-to-date is not the problem—there is plenty of AI research summaries available. The problem lies in the feasibility of using these methods. Can these findings be translated into acceptable results in practice?
To bridge the gap between AI research and applied AI, we created the AI Fellowship. The AI Fellowship is our internal research and development program for our machine learning engineers. You can read more about the AI Fellowship in this Medium article from Sybren Jansen, Head of AI Innovation.
The program is centred around four main objectives:
The AI Fellowship comprises four R&D cycles each year, with each cycle running for three months. It includes a research phase (7 weeks) and a development phase (2 weeks). The time left acts as a buffer. Whenever someone is sick, on a break, or wants to document their findings, the buffer can be used. And for each cycle, different topics are chosen. These topics often align with our engineers' interests and focus areas.
The end product of a cycle is the creation of technical content, typically in the form of blog posts or research papers. In the last year, our engineers have written about Graph Neural Networks, Biomedical Named Entity Recognition (BioNER), Transformers, and more. You can find these technical posts (and more) in our Medium publication.
Another form of output could be an open-source software package. We like to create open-source software so that others can benefit from our work. You can find our open-source repositories on our GitHub.
Curiosity and having an experimentation mindset are core traits here at Slimmer AI. If you’re curious about the AI Fellowship and interested in working with us, check out our open roles or reach out to our team.