Sign in

Tomasz Kalinowski

@t-kalinowski.bsky.social
2.2K followers 1.2K following 64 posts

Software Engineer at Posit PBC. I mostly post about R, Python, and Deep Learning. Github: github.com/t-kalinowski

PostsRepliesMedia
Reposted by Tomasz Kalinowski
Gabor Csardi @gaborcsardi.org · 14/09/2026
New beta of #Rstats rig: LOTS of EXCITING stuff: - new: web site: rig.r-lib.org - new: user-mode R installation: rig.r-lib.org/admin-vs-use... - new: manage R packages (CRAN + git): rig.r-lib.org/pkg-guide.html - new: manage R projects and virtual environments: rig.r-lib.org/proj-guide.h...
rig.r-lib.org
rig – the R Installation Manager – rig
35019
Tomasz Kalinowski @t-kalinowski.bsky.social · 02/09/2026
yaml12 0.2.0 is out for R and Python! The packages now write YAML that is much more human-readable: - automatic line wrapping for folded string blocks - fewer unnecessary quotes - literal blocks gracefully handle indentation R: posit-dev.github.io/r-yaml12/ Python: posit-dev.github.io/py-yaml12/
posit-dev.github.io
Fast YAML 1.2 Parser and Formatter
A fast, correct, safe, and ergonomic YAML 1.2 parser and generator written in Rust. Convert between YAML and simple R objects with full support for multi-document streams, tags, anchors, and aliases. ...
0163
Reposted by Tomasz Kalinowski
Posit @posit.co · 31/08/2026
Announcing cuda​.ml 0.4.0: GPU-accelerated machine learning in #RStats! Bring #NVIDIA compute into tidymodels & parsnip workflows with 2 commands. • Accelerated XGBoost & LightGBM inference via nvForest • Portable model persistence Read more: opensource.posit.co/blog/2026-08...
opensource.posit.co
cuda.ml 0.4.0: GPU-accelerated machine learning from R
cuda.ml 0.4.0 simplifies installation, expands tidymodels integration and tree-ensemble inference, and adds portable persistence for supported fitted models.
0268
Tomasz Kalinowski @t-kalinowski.bsky.social · 21/08/2026
{cuda.ml} v0.4 is out! It brings GPU-accelerated machine learning with NVIDIA RAPIDS cuML to R. It now handles CUDA/RAPIDS setup automatically—no manual CUDA or RAPIDS installation needed. It also adds first-class tidymodels support. opensource.posit.co/blog/2026-08...
opensource.posit.co
cuda.ml 0.4.0: GPU-accelerated machine learning from R
cuda.ml 0.4.0 simplifies installation, expands tidymodels integration and tree-ensemble inference, and adds portable persistence for supported fitted models.
0224
Tomasz Kalinowski @t-kalinowski.bsky.social · 18/08/2026
It normalizes, always into the most human-readable and compact form. Check out t-kalinowski.github.io/yamark/examp...
t-kalinowski.github.io
Examples – Yamark
Before-and-after examples of Yamark’s YAML, Markdown, and embedded formatting.
000
Tomasz Kalinowski @t-kalinowski.bsky.social · 18/08/2026
I've been working in Markdown and YAML a lot lately, and ended up building Yamark, a fast formatter for both. With LLMs, Markdown + YAML feel like programming languages. I wanted the same kind of tooling for them that I'm used to with code. t-kalinowski.github.io/yamark/
t-kalinowski.github.io
Yamark
A fast formatter for YAML and Markdown.
3243
Tomasz Kalinowski @t-kalinowski.bsky.social · 12/08/2026
Ask the llm to review old session transcripts, collect all the feedback you've ever given it about prose, and write a 'eugene-prose' skill, that distills what you do and don't like. Then tell it to always use that skill anytime it wants to present prose to you.
020
Tomasz Kalinowski @t-kalinowski.bsky.social · 06/08/2026
{cuda.ml} 0.4.0 is ready for CRAN, but CRAN is on break for two weeks. If anyone on Linux or WSL2 with an NVIDIA GPU has time to try the development version with the new `cuda_ml_install()` setup, I’d appreciate hearing how it goes. Feedback on the docs is welcome too: mlverse.github.io/cuda.ml/
mlverse.github.io
cuda.ml: R interface for RAPIDS cuML
An R interface to GPU-accelerated machine learning algorithms from RAPIDS cuML, powered by CUDA.
010
Tomasz Kalinowski @t-kalinowski.bsky.social · 24/07/2026
Yes! All the frontmatter fields have a command line counterpart, including --with, --exclude-newer, --r-version. `ir run --help` shows some examples: r-lib.github.io/ir/reference...
r-lib.github.io
Reference – ir
020
Tomasz Kalinowski @t-kalinowski.bsky.social · 24/07/2026
The next version of rig will allow ir to download and install R on-demand into a cached location :) Then R won't be a prerequisite, just another internal implementation detail.
020
Tomasz Kalinowski @t-kalinowski.bsky.social · 24/07/2026
Not yet, that's a good idea! The only prerequisite today is that R needs to be installed. (reticulate is an internal implementation detail).
220
Tomasz Kalinowski @t-kalinowski.bsky.social · 23/07/2026
I'm happy to share `ir`, a new command-line tool for running portable R scripts and Quarto documents. It's inspired by my two favorite parts of `uv`: self-describing scripts and tools you can run without installing first. opensource.posit.co/blog/2026-07...
opensource.posit.co
ir 0.1.0: self-describing R scripts and Quarto documents
`ir` is a new command-line tool for running portable R scripts and rendering Quarto documents whose package requirements, and optional R selection, live inside the file itself.
47521
Tomasz Kalinowski @t-kalinowski.bsky.social · 08/07/2026
I’m at useR! in Warsaw this week and talking about mcp-repl, an MCP server that gives agents sandboxed Python and R sessions: github.com/posit-dev/mc... Come say hi if you want to chat about AI tooling, deep learning in R — keras3, torch, jax, r-xla — or, of course, S7!
141
Reposted by Tomasz Kalinowski
Ambassador Frank Hull @frankiethull.bsky.social · 01/06/2026
@t-kalinowski.bsky.social saved me all the headaches mentioning py_require a few weeks ago. It saves so much time & works great 💯
131
Tomasz Kalinowski @t-kalinowski.bsky.social · 22/05/2026
All json is valid yaml too!
040
Reposted by Tomasz Kalinowski
Richard Iannone @richmeister.bsky.social · 21/05/2026
YAML deserves a defense! opensource.posit.co/blog/2026-05... The complaints are real but historical as they describe YAML 1.1, not the 1.2 spec that fixed them in 2009. The post traces the full history and introduces py-yaml12, a fast Rust-backed Python library for the modern spec. Hope you enjoy!
opensource.posit.co
In Defense of YAML
A common refrain among developers: YAML is bad and TOML is good. This post argues otherwise, tracing the history of configuration formats, examining what YAML 1.2 actually fixed, and introducing py-ya...
2225
Reposted by Tomasz Kalinowski
Daniel Falbel @dfalbel.bsky.social · 30/04/2026
Just published a new blogpost on torch for #rstats updates: opensource.posit.co/blog/2026-04...
opensource.posit.co
torch Ecosystem Updates
An overview of recent updates across the torch ecosystem.
0213
Reposted by Tomasz Kalinowski
Thomas Lin Pedersen @thomasp85.com · 20/04/2026
I am excited beyond description to lift the veil on what we have been working on in 2026: Please meet ggsql! A new extension of the SQL language for creating visualisations using the grammar of graphics. Read all about it in the blog post or visit the website at ggsql.org
opensource.posit.co
ggsql: A grammar of graphics for SQL
Introducing ggsql, a grammar of graphics for SQL that lets you describe visualizations directly inside SQL queries.
1338886
Reposted by Tomasz Kalinowski
Daniel Falbel @dfalbel.bsky.social · 15/04/2026
We just released raghilda to PyPi. Raghilda is a framework for building RAG systems in #python. See the release blogpost: opensource.posit.co/blog/2026-04...
opensource.posit.co
RAG with raghilda
An introduction to building retrieval-augmented generation systems with raghilda.
1185
Reposted by Tomasz Kalinowski
Richard Iannone @richmeister.bsky.social · 15/04/2026
We just released a blog post about something I'm very excited about: Great Docs. Post is in the *new* Posit Open Source site: opensource.posit.co/blog/2026-04... Great Docs is a #python pkg for creating beautiful Python package websites. We make it easy to use but also give you lots of options.
opensource.posit.co
Introducing Great Docs: Beautiful Documentation for Python Packages
Great Docs is a new documentation site generator for Python packages. It auto-discovers your API, builds a modern site with Quarto, and ships with LLM-friendly features and one-command deployment.
24215
Reposted by Tomasz Kalinowski
Dan Oehm @danoehm.bsky.social · 03/04/2026
Using {reticulate} and referencing R objects in Python with r.* and Python objects in R with py$* feels like God mode. #Rstats #Python
0141
Reposted by Tomasz Kalinowski
Max kuhn @topepo.bsky.social · 21/03/2026
Our list of 2026 #rstats and #python summer internships has been posted. We can't wait to work with you and make great things! tidyverse.org/blog/2026/03...
tidyverse.org
2026 Posit Internships
Posit is sponsoring four summer internship positions in 2026.
12618
Reposted by Tomasz Kalinowski
Emily Riederer @emilyriederer.bsky.social · 21/03/2026
Anyone else spoiled by R's pkgdown and find python's mkdocs or sphinx a bit tedious and config-heavy. Check out @richmeister.bsky.social 's awesome new {great-docs} as a batteries-included alternative to spin up an effective docs site in <15 min! 1/n 🧵 github.com/posit-dev/gr...
github.com
GitHub - posit-dev/great-docs: Great Docs lets you easily build a Python package docs site
Great Docs lets you easily build a Python package docs site - posit-dev/great-docs
34812
Reposted by Tomasz Kalinowski
Posit @posit.co · 06/03/2026
Meet Posit AI. Built for data scientists who want AI inside their RStudio workflow. Instead of black-box suggestions, Posit AI helps generate code you can inspect, understand, and rerun. Learn more: posit.co/products/ai?...
Graphic with Posit AI branding over a background of R code layered on an aerial forest landscape. Large headline text reads: “AI, embedded in RStudio. No more context switching.”
2245
Reposted by Tomasz Kalinowski
Charlie Gao @shikokuchuo.net · 06/03/2026
New: mirai skill for AI coding agents 🤖 Use with Claude Code/Codex/Cursor for writing async/parallel R code. Covers explicit dependency passing, daemon setup, mirai_map, Shiny integration ++ Install: npx skills add posit-dev/skills --skill mirai github.com/shikokuchuo/mirai/tree/main/dev/skills
github.com
mirai/dev/skills at main · r-lib/mirai
Minimalist Async Evaluation Framework for R. Contribute to r-lib/mirai development by creating an account on GitHub.
1113
Tomasz Kalinowski @t-kalinowski.bsky.social · 18/02/2026
I’m happy to announce Rapp v0.3.0 — a package that makes it easy to build and share polished CLI applications written in R. Read more: www.tidyverse.org/blog/2026/02...
tidyverse.org
Rapp 0.3.0
Rapp is an R front-end (like Rscript) that turns simple scripts into polished CLIs, with automatic argument parsing, generated help, and support for commands and installable launchers.
0427
Tomasz Kalinowski @t-kalinowski.bsky.social · 16/02/2026
Should be fixed now! Reticulate needed some internal updates for pandas 3.
010
Reposted by Tomasz Kalinowski
Posit @posit.co · 04/02/2026
Announcing yaml12: High-speed YAML 1.2 for #RStats & #Python 🐫 Built entirely in Rust, get up to 2x faster in R and 50x faster in Python vs. defaults, tag evaluation (like code execution) is opt-in, and strict YAML 1.2 compliance for consistent data. Learn more: tidyverse.org/blog/2026/01...
A graphic for Posit featuring a hexagonal sticker with a camel carrying boxes and "yaml12" text, over a blue wave background.
0224
Tomasz Kalinowski @t-kalinowski.bsky.social · 03/02/2026
It really varies by institution and PI, everyone has their own slight spin on it. This seems like a pretty good starting point that covers the basics well (at least, for a wet-lab context) biotechnology.rutgers.edu/sites/defaul...
biotechnology.rutgers.edu
040
Tomasz Kalinowski @t-kalinowski.bsky.social · 28/01/2026
quickr recently gained support for OpenMP! Just add a `declare(parallel())` above a `for` or `sapply()`! github.com/t-kalinowski...
171
Tomasz Kalinowski @t-kalinowski.bsky.social · 27/01/2026
ragnar 0.3.0 is now on CRAN! 🚀 An R toolkit for RAG from your docs: faster ingestion, better retrieval, Azure OpenAI + Snowflake embeddings, and MCP serving. tidyverse.org/blog/2026/01... #rstats #rag #ai
tidyverse.org
ragnar 0.3.0
The new release of ragnar adds faster ingestion, new embedding providers, improved retrieval, and new integrations for using ragnar stores from tools.
1309
Tomasz Kalinowski @t-kalinowski.bsky.social · 21/01/2026
Sounds like you’re describing querychat github.com/posit-dev/qu...
github.com
GitHub - posit-dev/querychat: Chat with your data in R and Python
Chat with your data in R and Python. Contribute to posit-dev/querychat development by creating an account on GitHub.
083
Tomasz Kalinowski @t-kalinowski.bsky.social · 20/01/2026
One pattern I've seen that intuitively makes sense: first ask the llm to generate a draft response, *then* ask it to retrieve documents and fact-check itself, ask if it needs to revise the answer in light of new info, then deliver the final answer.
120
Tomasz Kalinowski @t-kalinowski.bsky.social · 19/01/2026
I agree “looking at code” is squishy. Better code quality is an inevitable byproduct of diligent review, but the key part IMO is it is about continuously reality checking the produced artifact against your intent. It’s the only reliable way to detect when things are going off the rails.
010
Tomasz Kalinowski @t-kalinowski.bsky.social · 19/01/2026
When he (proudly!) stopped looking at the code.
110
Tomasz Kalinowski @t-kalinowski.bsky.social · 19/01/2026
I think it’s when you stop looking at and making effort to understand (and improve!) the code that things go off the rails past a certain size. You have to pay attention and care!
110
Tomasz Kalinowski @t-kalinowski.bsky.social · 08/01/2026
Yes! Performance is very similar between the R and Python interfaces!
020
Tomasz Kalinowski @t-kalinowski.bsky.social · 08/01/2026
Not sure, I’d have to profile it to know. My first guess is that the way extendr protects allocated objects adds some overhead, and if that’s the case, we could probably find a better approach in yaml12.
000
Reposted by Tomasz Kalinowski
Jumping Rivers @jumpingrivers.com · 08/01/2026
Hallucinations are a key challenge when using large language models in production. Our latest blog post shows how Retrieval-Augmented Generation (RAG) helps, with a practical R example using open-source tools. Read more on the Jumping Rivers blog: www.jumpingrivers.com/blog/retriev...
jumpingrivers.com
Retrieval-Augmented Generation: Setting up a Knowledge Store in R
Learn about retrieval-augmented generation (RAG) workflows for large language models. This blog provides a hands-on introduction, including example code for creating a working knowledge store using op...
0102
Tomasz Kalinowski @t-kalinowski.bsky.social · 07/01/2026
Some highlights: - 100% passing on the comprehensive YAML test suite: github.com/yaml/yaml-te... - In simple-use benchmarks, compared with the most commonly used packages, yaml12 is ~2× faster in R and >=50× faster in Python: posit-dev.github.io/r-yaml12/art... posit-dev.github.io/py-yaml12/be...
110
Tomasz Kalinowski @t-kalinowski.bsky.social · 07/01/2026
Announcing two new packages: `yaml12` for R and `py-yaml12` for Python. Fast, correct, and safe YAML 1.2 parsing and emitting, built in Rust. R: posit-dev.github.io/r-yaml12/ Python: posit-dev.github.io/py-yaml12/ #yaml #rstats #python
posit-dev.github.io
Fast YAML 1.2 Parser and Formatter
A fast, correct, safe, and ergonomic YAML 1.2 parser and generator written in Rust. Convert between YAML and simple R objects with full support for multi-document streams, tags, anchors, and aliases. ...
2264
Reposted by Tomasz Kalinowski
Sharon Machlis @smachlis.bsky.social · 16/12/2025
Turn YAML into R and R into YAML with the new {yaml12} #RStats 📦 posit-dev.github.io/r-yaml12/ind... By @t-kalinowski.bsky.social On the 📦 website: "YAML in 2 Minutes: A Gentle Introduction for R Users" posit-dev.github.io/r-yaml12/art...
posit-dev.github.io
Fast YAML 1.2 Parser and Formatter
A fast, correct, safe, and ergonomic YAML 1.2 parser and generator written in Rust. Convert between YAML and simple R objects with full support for multi-document streams, tags, anchors, and aliases. ...
173
Reposted by Tomasz Kalinowski
Barret Schloerke @schloerke.com · 10/12/2025
Shiny for R v1.12 is here with built-in OpenTelemetry support! 🎉 Trace sessions and reactive calculations in production with zero code changes. Just install {otel}/{otelsdk} and set your env vars. Learn more: shiny.posit.co/blog/posts/s... #RStats #RShiny #OpenTelemetry
0268
Reposted by Tomasz Kalinowski
Max kuhn @topepo.bsky.social · 12/11/2025
We're hiring an open-source #python developer focused on modeling APIs! tidyverse.org/blog/2025/11... #numpy #scipy #scikitlearn
tidyverse.org
Python Open-Source Developer
Posit is hiring a Python open-source developer to create more data analysis tools.
03414
Tomasz Kalinowski @t-kalinowski.bsky.social · 12/11/2025
I updated the gist with comments! In short: it configures the container with everything needed to let the agent run `devtools::test()` for the local R package within the container setup time limit. It installs R, configures CRAN, installs pkg dependencies, sets some env vars, etc.
110
Tomasz Kalinowski @t-kalinowski.bsky.social · 12/11/2025
I'd love to know what we can do to make this better! We've put a lot of work into making environment discovery "just work" with `py_require()` recently, but if you're still running into issues then our work is not done!
110
Tomasz Kalinowski @t-kalinowski.bsky.social · 25/10/2025
I think it will probably be straightforward to add support for GIL-free builds of Python, but it needs a little investigation at the moment. There is an open issue: github.com/rstudio/reti... Contributions welcome!
github.com
Add support for free threaded Python · Issue #1847 · rstudio/reticulate
Pointing reticulate at a freethreaded Python build currently raises an error. We need to investigate, diagnose the issue, and fix it. We should also make it easy to select a freethreaded Python in ...
110
Tomasz Kalinowski @t-kalinowski.bsky.social · 25/10/2025
reticulate v1.44 is on CRAN! Positron integration is now enabled by default. Enjoy R and Python REPL consoles powered by concurrent threads in the same process, zero‑copy R↔Python data transfer, and full Variables Pane support. Full changelog: github.com/rstudio/reti... #rstats #python #pydata
github.com
24911
Tomasz Kalinowski @t-kalinowski.bsky.social · 24/10/2025
Another option is to distribute it via github as a release asset. This would download the store at first use instead of at package install. Here is an example: github.com/t-kalinowski...
github.com
quartohelp/R/store.r at main · t-kalinowski/quartohelp
A ragnar-powered application. Contribute to t-kalinowski/quartohelp development by creating an account on GitHub.
110
Reposted by Tomasz Kalinowski
Umair Durrani @transport-talk.bsky.social · 14/10/2025
Work-in-progress #RStats package {brandthis} that enables you to efficiently create themes and color scales from personal/company branding. Uses {ragnar}, {ellmer}, {ggsci} and other R packages.
0131