The next "faster than Pandas" Polars interview will be out in 2 weeks, Part 1 came out yesterday, and recently I interviewed @theNeomatrix369 Kaggle Competition Expert (2 parts) on his favourite ML/DS tools - see the archives here:

At last week's Leadership session I also handed out snake stickers so folk could self-identify after the session to continue network, this feels like a nice iteration on previous sessions (where I didn't have stickers!), write-up:

Leadership discussion participants at #pydatalondon last weekend, thanks all for brilliant discussion, write-up:

Leadership discussion voting results at #pydatalondon last weekend, qs from DS team-leaders in the room, write-up:

Lots of fun, thanks all (even if I did have to miss the meal!).

RT @cheukting_ho: Thank you all these wonderful volunteers for making #PyDataLondon another success this year

RT @arnicas: See you there!.

RT @pydatalondon: 📣 16 Days Left for PyData London 2024 ! 🥳. 🎟️ GET YOUR TICKETS NOW 👉 🚨 LIMITED RELEASE (First Co….

Next newsletter: "what questions do you wish you'd been asked sooner?" from my RebelAI leadership group, make Pandas 2.x faster with CopyOnWrite, notes on the upcoming PyDataLondon conference in 2 weeks:

I wrote another (rather intermittent of late!) NaN newsletter, this one on my RebelAI Leadership group and better hiring processes plus a way to get Pandas 1.5+ to join faster than you might expect:

For the PyDataLondon 2024 conference on the Saturday morning (June 15th) I'm running another of my regular Leaders sessions, circa 60 leaders attend each year, we solve issues and swap good process, see here: (reply for GCal).

PyDataLondon Conf 2024 is coming up on June 14-16 (fri-sun), the schedule is excellent and tickets are available, same venue as last year and 10 YEARS OLD!:

New newsletter - best tips from my recent Pandas2/Polars/Dask talk, forecasting on Metaculus and making a Polars plugin:

This week's newsletter - Python 3.13 forthcoming copy-and-patch JIT, value and brand cost of LLM chatbots, pandas 2.2:

To help with llama (with GPU offload) testing I'm playing with the excellent nvtop: - see GPU% and GPU-mem% in the console, plus the processes running on it:

I gave a short talk @pydatalondon on Tuesday on llama.cpp running local LLMs on your CPU (no data sent off the machine), with live demos including multimodal with llava - slides here with cmd lines:

RT @John_Sandall: It's going to be an exciting year ahead for @pydatalondon. Changes are afoot, new venues, new sponsors, a growing team of….

RT @pydatalondon: PyData London co-founder and lightning speaker @ianozsvald introduces running llama.cpp on CPU at ⚡️speed. "It just work….

New newsletter - some lessons from my RebelAI leadership community on building a backlog & setting up DS for success plus 4 useful open src tools:

Newsletter - VizTracer to understand a multiprocess Python simulator's behaviour, the 1 feature in Python 3.12 I rather like (not type hints), DNA sequencing of sourdough cultures and a new Polars extn:

Latest newsletter - ChatGPT for unit testing during my class, Polars with better error messages leads to faster timeseries code development, my RebelAI peer leadership group is up and I share some first objectives: