Ned Letcher

Ned Letcher

@ned.sh

Data science, AI/ML, analytics, visualisation. Naarm/Melbourne @thoughtworks, #dataBS, #Python, #NLP, #DuckDB, & assorted whimsical miscellania

3.6K followers9.9K following135 posts

Pinned

Heard the good word about #DuckDB? Wondering how to use it to accelerate your data science/analytics & data/software engineering projects? Getting Started with DuckDB might be for you! Simon & I have made a practical guide full of hands-on examples, incl both #python & #rstats packt.link/byKYt

Just a few hours left to submit to PyCon AU! If you have a #python and/or #Data and #AI idea you're keen to share with the community, why not consider a cheeky last minute submission? 🐍

PyCon AU@pyconau.bsky.social

There are just a few hours left to submit your talk, workshop and poster ideas to PyCon AU 2026! Everything you need is at 2026.pycon.org.au/cfp The deadline is '29 March Anywhere on Earth' - thats later tonight for everyone Australia and New Zealand.

Ned Letcher reposted

I'm headed to San Diego this afternoon to attend the mech interp workshop at #NeurIPS! Really excited to give a lightning talk and catch up with everyone. 🌴

I'm in town for NeurlPS all week. Wow this conference is biiiig. Anyone I know (or don't know!) up for a drink? #NeurIPS #NeurIPS2025 #mlsky

Ned Letcher reposted

I like to think that this map of all the regional names for woodlice in the UK could also be a map illustrating the new tribes of humans to be found on the same landmass after The Great Digital Crash and ensuing breakdown of society.

Ned Letcher reposted

The videos from the #DataBS Conference have made it to YouTube! Over the coming days, we'll highlight sessisons from the event -- one post per talk. You can also sneak ahead by going straight to the YouTube playlist, here: www.youtube.com/playlist?lis...

youtube.comData Behind the Scenes 2025 - YouTubeData Behind the Scenes is a data conference by data practitioners, for data practitioners. We come together to share stories and experiences about the the me...

Ned Letcher reposted

It's been a week since we launched @chartle.cc with @adnaanj.bsky.social and we had an amazing start! - Average of 700 users/day - Record 1.2K users yesterday - From 78 countries - 66% win overall - 2m 18s average engagement Thank you to all who played + shared the game Any feedback, send our way!

Erwan Rivault@erwanrivault.com

Launch day 🚀 We’ve just released @chartlecc.bsky.social - a daily chart game! Your job is to guess which country is represented by the red line in today's chart. You get 5 tries, no other clues! Play today, come back tomorrow for a different chart with new data and share with your chart friends 📈

Ned Letcher reposted

📈 DuckDB 1.4.0 is out! This is our first LTS release which comes with *one year of community support*. It also supports database encryption, the MERGE SQL statement and Iceberg writes. For more details, read the announcement blog post at duckdb.org/2025/09/16/a...

Assessing LLM output quality requires managing their vast input/output spaces and acute sensitivity to minor prompt changes, but many teams are still running on largely vibe-based evals, incurring considerable risk and preventing systematic product improvement. We can do better! #databs #mlsky

View full thread

The course validated my intuitions that these foundations apply just as much to generative AI applications and has given me a road-tested pragmatic framework for applying them to the weird world of LLMs and AI engineering.

Assessing LLM output quality requires managing their vast input/output spaces and acute sensitivity to minor prompt changes, but many teams are still running on largely vibe-based evals, incurring considerable risk and preventing systematic product improvement. We can do better! #databs #mlsky

Their approach thoroughly resonates with my experience as a data scientist: look at your data, perform thorough error analysis, develop metrics that matter, and iterate via experimentation and user-feedback.

Ned Letcher reposted

There is a reason why the introduction and first three chapters are required reading in weeks two and three of my law and technology course Essential for understanding where we are

Faine Greenwood@faineg.com

I finally read computer scientist Joseph Weizenbaum’s 1976 classic “Computer Power and Human Reason.” This book deserves a massive revival in our current age of grotesque and largely thoughtless AI creep into everything:

Ned Letcher reposted

DuckDB is awesome and it’s a great tool to explore and transform data. But DuckDB doesn’t help you visualize and share data with others. That's why I built Shaper: taleshape.com/blog/turn-yo... #duckdb #databs #dataengineering

Ned Letcher reposted

Data Behind the Scenes Conference free registration is open! Wed Sept 24, ~10am-5pm EDT. Online-only, single track. Join data folks as they share experiences with how data work is done in production, thorns & disturbing hacks and all! (Pls help get the word out 📈) ti.to/databsconf/d... #dataBS

ti.toDataBS Conf"Data, Behind the Scenes" is a free-to-attend online-only, single track conference centered on the real stories of data work from the folks in the trenches. We’re not here for the latest AI hype, perf...

Ned Letcher reposted

🚀 We've just open-sourced Embedding Atlas – a tool for exploring large embedding spaces through rich, interactive visualizations 📊.

Ned Letcher reposted

I’ve been working on an experimental, multiplayer, open-world AI chat system called Numinex. 1. Uses branching comment trees 2. Multiplayer/multimodel 3. Explicit context curation 4. Built on ATProto 5. Open observation of model behaviors

Ned Letcher reposted

Ned Letcher reposted

Ned Letcher reposted

I have an opportunity to hire a staff scientist for my lab. Looking for someone with outstanding skillset in ML/statistics, genomics applications; interest in mentoring, strong publication record, PD experience required. Email CV to me+cc my assistant (see 'contact' on my website). Ad to follow.

Ned Letcher reposted

the peak moment of my career was co-inventing (with Roman Garnett) Bayesian optimisation, only to discover just before we submitted that someone had already invented it a few decades before

Ned Letcher reposted

After all the discourse over whether super-intelligent, ultra-powerful AI is coming, we now have a prime example of exactly how AI is massively disruptive, today: When it is used by lazy, reckless, and malicious people in power. Welcome to the age of AI-generated economic collapse, everyone

bloodinthemachine.comWelcome to the age of AI-generated economic collapseOn AGI, AI-generated tariffs, and what really threatens the social order

Ned Letcher reposted

Thoughtworks have published their latest #TWTechRadar. A few 'blips' (as they call them) of note in the data space: 🟢 Adopt: Data product thinking 🟢 Adopt: Trino 👍 Trial: Databricks Delta Live Tables 👍 Trial: @metabase.com ✋ Hold: Reverse ETL www.thoughtworks.com/radar #dataBS

thoughtworks.comTechnology Radar | Guide to technology landscapeThe Technology Radar is an opinionated guide to today's technology landscape. Read the latest here.

This is a thing that seems to keep recurring for me too. Any #databs peeps have any ideas?

Looking for a lightweight and serverless framework to manage transforms between OLAP assets in S3. Currently looking at pandas/duckdb in lambda using step functions - which works, but there is a lot of boilerplate and plumbing. Surely, this is a semi-solved problem?

Ned Letcher reposted

If you're using duckdb in a python script or jupyter notebook, you can run con.execute('CALL start_ui()') at any point, and the ui will pop right up in your web browser with the current database automatically available. (I knew about the UI, but I had missed this trick!)

Ned Letcher reposted

Are you training self-supervised/foundation models, and worried if they are learning good representations? We got you covered! 💪 🦖Introducing Reptrix, a #Python library to evaluate representation quality metrics for neural nets: github.com/BARL-SSL/rep... 🧵👇[1/6] #DeepLearning

Ned Letcher reposted

I was interviewed by @attacus.net for last night's broadcast of Byte into IT on RRR Melbourne Independent Radio. This is such a great conversation about my new book! Honestly one of my favorites so far. Huge thanks to @attacus.net. Listen to the episode here. www.rrr.org.au/explore/prog...

rrr.org.auPrograms: Byte Into IT – 19 March 2025, Byte Into IT — Triple R 102.7FM, Melbourne Independent RadioAn episode of Byte Into IT on 19 March 2025