Jacek Laskowski
books.japila.pl
Jacek Laskowski
@books.japila.pl
Freelance Data(bricks) Engineer • Databricks MVP • #ApacheSpark #MLflow #DeltaLake #UnityCatalog #Databricks #Python #uv • Java Champion • ASF member • he/him

Into #LLMOps #MLOps #GenAI

https://www.linkedin.com/in/jaceklaskowski
Thanks @waitingforcode.bsky.social for the book! 👏👏👏

I'd read a few chapters already while in writing, but a hardcopy is so much better experience! 💯💖📚

#books #reading #DataEngineering
June 6, 2025 at 4:04 PM
In #Databricks ML (a hosted #MLflow), models are considered functions 👍 Makes a lot of sense IMHO.

Interestingly, there's another way to refer to a model.

As a "routine"! 🤯

In Databricks CLI, --force argument is only supported for functions (not model-versions) 😉
May 18, 2025 at 4:21 PM
Got the whole stack working! The latest versions of #ApacheSpark 4.0.0 RC6, #UnityCatalog 0.3.0-SNAPSHOT and #DeltaLake 3.4.0-SNAPSHOT all work together! 🎉
May 16, 2025 at 6:50 PM
Don't remember when/if I ordered the book, but it's worth my time so far 📖💯 The first 10 pages were very engaging ❤️

Designing LLMs Applications from @oreilly.bsky.social 👍

#books #reading #LLM
May 14, 2025 at 9:43 PM
Productivity boost arrived! 🏆

Building #UnityCatalog OSS took...1s! 🤯

Next up are #ApacheSpark 4.0.0 RC4 and #DeltaLake 3.4.0-SNAPSHOT

#macmini #m4pro
May 1, 2025 at 6:14 PM
New #ApacheSpark 4.0.0 RC4 cut this morning ❤️

github.com/apache/spark...
April 10, 2025 at 1:36 PM
#pydantic 2.11.0 hit the shelves, and just before I was about to delete the email with the news I found this... 🤩

I completely forgot I made my contribution to the lib! 🤪

github.com/pydantic/pyd...
March 28, 2025 at 9:00 PM
Yay! Another OSS repo my PR went to 🥳

It's yours truly moi in #mlflow repo, baby! 😎

➡️ github.com/mlflow/mlflo...

BTW, mlflow v2.20.3 went out. No idea how important it is as I'm just getting started... 🤷‍♂️

Do you use it? How? When? What for? LMK 🙏
February 26, 2025 at 10:21 PM
It appears that a brand new #DeltaLiveTables pipeline created by #DatabricksCLI as a Databricks Asset Bundles (DAB) project is no longer legit 🤨

It may be due to LIVE schema marked as legacy, but it does not seem to be used so investigating... 🧐

➡️ docs.databricks.com/aws/en/delta...

#Databricks
February 23, 2025 at 12:24 PM
Thank you, @jetbrains.com 🙇‍♂️🥂

#PyCharm #Python #IntelliJIDEA #Scala
February 16, 2025 at 2:17 PM
Just a tiny contribution to #pydantic, such a huge excitement! 😎

➡️ github.com/pydantic/pyd...

#Python #OSS
February 2, 2025 at 12:39 PM
I missed the day when this tiny PR went in to #UnityCatalog repo, but I'm not certainly gonna miss this oppty to tell you about it, even a few days after 😆

➡️ github.com/unitycatalog...

That made me #13 in the contributors list of superc00L folks with 15 commits 😎

➡️ github.com/unitycatalog...
January 12, 2025 at 12:58 PM
One of the features of #DeltaLake 3.3.0 is IDENTITY Columns (table feature).

IDENTITY columns allow assigning unique values for each record written out into a table (unless column values are provided explicitly).

➡️ books.japila.pl/delta-lake-i...

More to come soon...
January 11, 2025 at 9:05 PM
📢 #DeltaLake 3.3.0 hit the shelves 🥳

➡️ github.com/delta-io/del...

Aaaaand, you can find some familiar names among the contributors 😎

/me Updating books.japila.pl/delta-lake-i... ...
January 10, 2025 at 9:47 PM
#parkrun tomorrow gonna be...snowy 🥶😱

My first winter #running season and I'm so scared of even thinking about low temperatures ☠️

At such low temps, even 00:25 looks challenging 😬
January 10, 2025 at 7:34 PM
#DeltaLake 3.3.0 is very very close 🥰

➡️ github.com/delta-io/del...
January 6, 2025 at 5:31 PM
Building #ApacheSpark 4.0.0.dev0 (master) takes 1h on mac mini / Apple M2 Pro 😱

Would M4 make it faster? Half the time, perhaps?
January 5, 2025 at 2:59 PM
There must've been a glitch in @bsky.app domain verification as mine been "Invalid Handle" for a while 😬

All's good now 😎 Just a few clicks and all green ✅

Wonder if that's had any impact on account discovery? How am I reached out if I'm "Invalid Handle"?
January 5, 2025 at 1:17 PM
One of my resolutions for 2025: champion #uv as the only #Python project manager of choice. It is so pleasant to use!

#ApacheSpark Connect requires some extra deps. No need for a venv, just "uv run --with".

Make uv your Python project manager in 2025 🙏

➡️ docs.astral.sh/uv/

HNY 🥂🥳
January 1, 2025 at 2:37 PM
Is this "Generative AI" warning callout OK with you? 🫣

➡️ books.japila.pl/spark-connec...

I'm sold using it more often as a way to improve my English writing skills ✍️ and, more importantly, improve » your « knowledge of #ApacheSpark 😜
December 31, 2024 at 4:12 PM
I'm far from being comfortable and even prepared to talk to AI Assistants / #LLMs in their language 😎

#PromptEngineering is an art, don't you think? 🤯

Slowly getting used to this way of working. Even fewer people to talk to 😱🤷‍♂️

BTW, how'd you prompt for a title for SparkSession.getOrCreate?
December 31, 2024 at 12:12 PM
I'm really, really surprised (almost shocked) how precisely this #LLM model explains the gist of a #Scala def. I don't think I could explain it better. Spot on!

I'm convinced that I should use LLMs more often (perhaps even too often 😬).

And it's all within #IntelliJIDEA ❤️
December 29, 2024 at 3:57 PM
Got curious about #Ollama and, while in the Github repo, I found that:

1️⃣ Ollama supports importing #GGUF models
2️⃣ #HuggingFace Hub supports all file formats, but has built-in features for GGUF format

Lots uncovered. Little time 🫣

➡️ ollama.com
➡️ github.com/ollama/ollam...
December 29, 2024 at 2:22 PM
So I can share a bit more about the book "Prompt Engineering for LLMs" 😍

First of all, I'm confident I'm gonna read it all 👍

Then, while I was in #IntelliJIDEA I turned on the AI Assistant and found this little gem = you can run #Ollama locally and plug it in to the IDE ❤️
December 29, 2024 at 2:22 PM
Merely on Chapter 2 of "Prompt Engineering for LLMs" book 📚 from #OReilly, and yet tons of A'ha moments to understand how to use LLMs in my work! 😎

More importantly, I simply had no idea how to use it for the many repetitive tasks of mine (like code reviews) 🤷‍♂️

➡️ www.oreilly.com/library/view...
December 29, 2024 at 2:22 PM