Skip to content

ML learnings with Svante Sörberg

Svante originally joined Modulai as a master’s thesis student and has been with the company for over two years, these are his takeaways so far.

ML learnings after two years at Modulai with Svante Sörberg

Svante originally joined Modulai as a master’s thesis student and has been with the company for over two years, these are his takeaways so far.

Looking back at your time at Modulai, what stands out to you the most? 

It’s hard to pick one thing, but something that definitely strikes me looking back is the learning culture. When I joined Modulai, the most significant factor in that decision was the feeling that it would be a great place to develop and learn new skills. I don’t think I could have hoped for something better in that aspect. People are just really excited to share ideas, talk about their work, papers they’ve read, problems they’re facing, and so on. Even if you’re working with one client, as I have for most of my time here, you learn so much from everything else we’re involved in.

Is the work as an ML engineer as expected?

I have realized that the software engineering aspect of the job is more significant than I initially thought. I thought that the focus would mainly be creating Machine Learning models. However, in reality, there is a lot of work required to integrate the model with the client’s tech stack and legacy system to deliver value. Each client has their own unique requirements, and sometimes, it can be challenging to work with or against their existing infrastructure.

What kind of ML projects have you been involved in over the years?

Recommender systems, sales & marketing optimization, and ML infrastructure mainly. More recently, applying LLMs to automate the mundane parts of peoples’ everyday work.

This past year has been quite a ride for machine learning and generative AI development. How has it impacted your work?

The impact of GenAI on my work has been twofold. Firstly, I have integrated it into my daily workflow, using tools like Github Copilot to speed up the creation of boilerplate code. Chat-based tools have also made it easier for me to work with things that were previously outside of my skillset, such as JavaScript and web development.

Secondly, GenAI has expanded the scope of what I work with. With pre-trained LLMs, the barrier to solving certain problems has been reduced immensely. This has allowed me to tackle old and stubborn business problems that previously weren’t worth the effort to solve.

What is the most important tip you would give to someone applying for a Master’s Thesis project at Modulai?

I think the master’s thesis is probably the first time in your academic career that you really become the expert on your topic. Don’t be afraid to challenge any theories or conclusions that don’t seem to make sense when you’re doing the literature review – there is a lot that doesn’t hold up to close scrutiny.

I have realized that the software engineering aspect of the job is more significant than I initially thought.

Svante Sörberg

Wanna read more about the projects Svante has been working with?