@przemur Profile picture

Przemek Maciolek

@przemur

I like solving riddles 🇵🇱🇪🇺🚴🏼‍♂️

Similar User
Marc Brooker photo

@MarcJBrooker

Łukasz Langa photo

@llanga

Peter Kowalczyk 🍍 photo

@peter_kow

Basia Madej-Romaniuk 🌻 photo

@basiamadej

Vincent D. Warmerdam photo

@fishnets88

e/la photo

@elamadej

Monika Starzyk photo

@stmonika

Karol Nowak (@knowak@astrodon.social) photo

@knowak

Przemek photo

@powczarek

Michał Bugno 💙💛 photo

@michalbugno

Can LLMs evaluate morality? Turns out, they’re surprisingly insightful. I tested political debates through the lens of @JonHaidt's Moral Foundations Theory. Check out my findings here: open.substack.com/pub/przemur/p/…


Przemek Maciolek Reposted

Hannes shared this over on the butterfly app, but if this DuckDB PR works, then table sampling is about to get wildly efficient! This could open up whole new avenues for visualizing the distributions of very large tables in our Column Explorer feature. github.com/duckdb/duckdb/…


Przemek Maciolek Reposted

LLMs are powerful sequence modeling tools! They not only can generate language, but also actions for playing video games, or numerical values for forecasting time series. Can we help LLMs better model these continuous "tokens"? Our answer: Fourier series! Let me explain… 🧵(1/n)


Przemek Maciolek Reposted

I'm sending e-mails to my investors every month. I started including three database trends I find intriguing. Here are my favourites from October 2024. 🧵1/4


🛠️ Ever wondered how to create an analytical database query engine from scratch? Join me on this journey as we explore how we implemented SOL at @MotifAnalytics to tackle complex event pattern matching via automatons #TechBlog #DatabaseEngineering motifanalytics.com/posts/how-to-b…


Przemek Maciolek Reposted

Does this career path diagram match your mental model? We've recently got our hands on a dataset of career paths in data science and analytics. Working on sequence analytics at Motif, of course, we couldn't resist calculating major career paths with promotion times and rates. It…

Tweet Image 1

Przemek Maciolek Reposted

1/ New paper @Nature! Discrepancy between human expectations of task difficulty and LLM errors harms reliability. In 2022, Ilya Sutskever @ilyasut predicted: "perhaps over time that discrepancy will diminish" (youtu.be/W-F7chPE9nU, min 61-64). We show this is *not* the case!

Tweet Image 1

Przemek Maciolek Reposted

One of @MotifAnalytics's goals is to show that it is possible to "pull data real quick" on practical user behavior questions and have fun doing it. Building such an analytics tool involved re-thinking the core analysis loop: - finding a better set of operations for working with…


Przemek Maciolek Reposted

Snowflake and Redshift have both published representative samples of real-world queries, and the way people actually use these systems is INTERESTING. The main workload is ingest and transformation!

Tweet Image 1

Przemek Maciolek Reposted

Flooding disaster unfolding right now in Central Europe. Why is it so bad? This thread takes a quick look at some of the key ingredients of this record-breaking storm.

Tweet Image 1
Tweet Image 2

Once trying a Progressive Analytics system, there's no turning back. We are proud to be part of this movement and built a high-throughput (millions of events/s), low-latency (<1s P50) and scalable (1-1000's of workers) system using TypeScript and @dragonflydbio Read on how


Przemek Maciolek Reposted

DuckDB now supports the Arrow PyCapsule Interface, which means you no longer need pyarrow to exchange Arrow data in python! Convert to/from @DataPolars (*in principle), DataFusion, pyogrio, Lonboard, arro3, all without pyarrow!

We are happy to release DuckDB v1.1.0 “Eatoni”. The new release packs a ton of new features: friendly SQL extensions, performance improvements and spatial features. It also includes improvements towards supporting Community Extensions. See our blog post: duckdb.org/2024/09/09/ann…

Tweet Image 1


Przemek Maciolek Reposted

So many amazing ergonomic improvements in this release of DuckDB. I'll break down a few in thread ~

We are happy to release DuckDB v1.1.0 “Eatoni”. The new release packs a ton of new features: friendly SQL extensions, performance improvements and spatial features. It also includes improvements towards supporting Community Extensions. See our blog post: duckdb.org/2024/09/09/ann…

Tweet Image 1


Przemek Maciolek Reposted

Where do folks who work in data roles stick around the longest? You can use @MotifAnalytics to find out quickly! There is an interesting company with the longest average tenure. We've ungated access to Motif so try it out and let us know what you learn🚀 loom.com/share/39e6efdf…


Przemek Maciolek Reposted

At @MotifAnalytics, we often hear from customers that they want to know what drives the successful and unsuccessful outcomes in their products. ✨Today we're announcing our Causal Discovery Engine for automated causal inference that scales to 1000s of cause-effect hypotheses✨


Przemek Maciolek Reposted

I heard the term "headless data architecture" for the first time a few months ago. Now I hear it everywhere. The idea is that you can "bring your own query engine," but you're using object storage, a table format, and a catalog.


Przemek Maciolek Reposted

It took three years to finish, but our follow-up to the 2006 "What Goes Around Comes Around" is finally out! Stonebraker and I examine the last 20 years in databases and discuss why relational databases + SQL will continue to remain on top. 📄PDF: db.cs.cmu.edu/papers/2024/wh…

Tweet Image 1

Przemek Maciolek Reposted

TL;DR: In the last 3 years, @duckdb has become 3-25x faster and can analyze ~10x larger datasets all on the same hardware. 🤯

New blog post by @__AlexMonahan__: Benchmarking Ourselves over Time at DuckDB The DuckDB team's philosophy is to first ensure correctness, then iterate and optimize to improve performance. This blog explores how this happened over the last three years, when DuckDB became…

Tweet Image 1


Loading...

Something went wrong.


Something went wrong.