Happy April!

As the sun is finally coming out and days are getting longer, it’s the most wonderful time to experience the creations of Mother Nature coming to life. If the weather is good, load this newsletter and the related articles into your favourite text-to-speech tool and head out for a walk :)

Meet the curator of the month

Rasmus is a Data Scientist specialising in Natural Language applications. His main interest lies in how computers can be used to enable people to focus on more meaningful things in life. Outside of work, Rasmus likes to lift heavy things at the gym, play golf and other ball games.

Rasmus' picks

When to fine-tune — and when not to

By Andrew Ng

Fine-tuning small language models is constantly gaining a bigger traction in solving real-world problems. However, Andrew highlights that as fine-tuning is still indeed very-hard and time-consuming to do, it should not be needed for most of the use cases, as improving prompts and agentic workflows provide good enough results with a much simpler solution architecture. For some use cases, fine-tuning is the critical last piece of the puzzle.

Learn more

Sustainable and efficient vector search solutions

By Elias Herranen

An interesting study for anyone interested in creating more resource-efficient Generative AI applications.

Learn more

Tracing the thoughts of a large language model

By Anthropic

Explores how concepts and reasoning steps can be traced layer by layer, explaining the model's decision-making process. Solid read for developers interested in understanding what’s happening behind the scenes when an LLM generates a response.

Learn more

Know your curator

How do you stay updated on the latest industry trends and technologies?

To try to keep up with fundamentals and avoid reading fuzz, I try to read stuff that would, in my mind, make it to a weekly or a monthly newspaper rather than articles that would only make it to the daily newspaper.

Are there any emerging technologies or trends you’re particularly excited about?
Historically, programming required a thorough understanding of computer hardware and low-level languages. Over time, we have abstracted many of those complexities away; first through higher-level languages, then IDEs, and now AI-assisted tools. This has lowered the barrier for developing (most) software significantly.

I think we’re heading toward a future where plumbers, lawyers, or teachers —professionals without formal software engineering backgrounds — will program small hacks to improve their day-to-day work. This shift would be powered by platforms and tools created by professional software engineers, and could unlock a new wave of innovation in how computers can be used.

Join us!

Data Scientist

If you're an accomplished Machine Learning engineer who is passionate about Generative AI, LLMs, and RAG, this is an excellent opportunity to join our innovative Data team who's actively working on our Generative AI initiatives.

Helsinki >> | Tampere >>

See all open positions globally

Also – don’t forget to share DevBreakfast with your friends and colleagues!

Find us on social media

LinkedIn | Facebook | Twitter | Instagram

Address source: Futurice’s client or interest group register | Privacy Policy