Just watched Martin Odersky keynote at ScalaDays 2025, about Scala future Capabilities, and he's pretty convincing.
It shows what could be the shape of Scala programming in the coming years, and what Scala 3 was aiming at.
(he was less convincing when talking about project migration from effect systems, because it's another way of functional programming)
Next up, a talk by Anna Herlihy of LAMP and DIAS about macro-free type-safe database queries.
Having refactored my fair share of bindings to NoSQL DBs, I am definitely looking forwards to understanding better how to create universal interfaces that cover both NoSQL query languages and SQL!
#ScalaDays #ScalaDays2025 #Scala
11/🧵
Interesting talk by Muayad Sayed Ali of WRITER on the GenAI in Entreprise:
1. Obstacles to GenML adoption in the company
2. Why use Scala in LLM serving stack rather than Python
#ScalaDays #ScalaDays2025 #Scala
10/🧵
In the afternoon, Kannupriya Kalra and
Rory Graves will talk about building the cool stuff in Scala using LLMs4S, focusing on building LLM applications based on Scala.
#ScalaDays #ScalaDays2025 #Scala
9/🧵
Immediate after me, Krzysztof Romanowski will give a talk on the relationship between Scala, AI and productivity, at all stages of a software project
#ScalaDays #ScalaDays2025 #Scala
8/🧵
Tomorrow morning, I will open the chain of AI x Scala talks right after the keynote, talking about our experience with making LLM-generated code safe and robust, how static verfication failed us, but Scala type system came to the rescue.
#ScalaDays #ScalaDays2025 #Scala
7/🧵
Today, Jakub Kozłowski and Michał Pawlik will talk about code generations that has nothing to do with LLMs, for a change, instead focusing on formal constraints and satisfying them, and what they have learned from trying to address real-world problems with them.
#ScalaDays #ScalaDays2025 #Scala
6/🧵
4. Failure to deliver by Scala teams. Perhaps a consequence of the #3, but I have on several occasions heard from technical managing positions about projects they initially hired Scala developers for data science projects, saw them failing to deliver on time; re-hired a Python/Rust team and saw them actually deliver.
While focus on success stories is great, good post-mortems are also essential and would be nice to see.
#ScalaDays #ScalaDays2025 #Scala
5/🧵
2. Scala is not that easy to pick up. A recurrent complaint from beginner programmers I hear is that Scala code is pretty scary at first, using a lot of unfamiliar keywords with unclear effects.
3. Lack for domain-specific libraries. Python is popular because no matter your domain, you will be able to find a library that does most of what you want to do, in the language you understand. numpy, statsmodels, scikits-learn, pytorch. not so much in Scala.
#ScalaDays #ScalaDays2025 #Scala
4/🧵
Intersting talk by Daniel Ciocîrlan of Rock the JVM on what made Scala Popular in 2010s, why its popularity dropped in 202s and who Scala can appeal to today.
My perspective as a person external to Scala is somewhat different.
1. Killer App. In 2010, smartphone wave arrived, and with it a need to write Android Java and Opjective-C by the same people. Coming from Objective-C, 2010 Java was not palatable, unlike Scala. Now Kotlin does the same thing.
#ScalaDays #ScalaDays2025 #Scala
3/🧵
Next upcoming ML x Scala talk, later this morning is by Olga Chuchuk and François Laroche of NuMund, presenting the Full-Stack Scala used at NuMind to train best-in-class structured information extraction LLMs.
#ScalaDays #ScalaDays2025 #Scala
2/🧵
Covering Scala Days'25, as Program Committee member focused on ML x Scala synergies.
#ScalaDays #ScalaDays2025 #Scala
1/🧵
