AI at Google/Kaggle (inversion). PhD chemical engineer. Too many hobbies.
Neenah, Wisconsin 54956
Joined December 2007
- Tweets8.3K
- Following2K
- Followers9.2K
- Likes10.4K
Walter Reade retweeted
📣 If you have opinions about chart crimes and plotslop, we have an arena for you. Go vote for the best models on Chart Arena, live on Kaggle now! kaggle.com/benchmarks/kaggle…
ALT A screenshot of Chart Arena on Kaggle showing plots created from prompts+data for side-by-side voting.
A critical skill for using agents effectively is the ability think in terms of root cause problem solving. Instead of telling an agent to "Stop doing <bad thing>" you tell it to, e.g., "Develop a system to detect and correct when <bad thing> happens".
Of course, there are even better variants of this. But the point is that people who think in terms of problem solving and countermeasures will get much more out of agents.
I expect some new optimal configurations to come from this.
This is a well-studied open problem in combinatorial geometry: the best-known solutions come from decades of work by mathematicians and programmers.
Join the competition here: kaggle.com/competitions/squa…
7
722
Walter Reade retweeted
The ideal outcome for science isn't a couple of labs using proprietary models and millions of dollars to crack trophy problems, it's to have millions of AI-aided human scientists who follow their curiosity and become way more ambitious
Walter Reade retweeted
Huge congratulations to Francis Halzen on the Nobel Prize, a landmark moment in astrophysics! 🏆
It was an honor for our Kaggle community to collaborate with the IceCube Neutrino Observatory in 2023, building open-source ML models to help predict the direction of neutrinos.
BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2026 #NobelPrize in Physics to Francis Halzen “for decisive contributions to the IceCube Neutrino Observatory and the discovery of high-energy neutrinos of astrophysical origin.”
Walter Reade retweeted
🏁 That's a wrap! The @biohub Cell Tracking challenge on @kaggle has ended.
4,478 participants, 3,947 teams, 120 countries, 95,438 submissions, tracking every cell of developing zebrafish embryos in 3D over time. Thank you! 🧵🐟
kaggle.com/competitions/bioh…
Walter Reade retweeted
I'm excited to announce CASMI 2026, the blind challenge hosted by @enveda on Kaggle to address one of the grandest challenges in life science: can you tell the structure of a small molecule from its mass spectrum?
kaggle.com/competitions/enve…
Walter Reade retweeted
My favorite thing about automating verifiability of many scientific topics is that we’ll FINALLY be able to get rid of “peer review” once and for all. What a horrible, horrible, system that is.
Walter Reade retweeted
A sister's team is hiring at NVIDIA. @kaggle competition GM are welcomed, but not only. Location can also be Europe even if the description says US.
jobs.nvidia.com/careers/job/…
I helped beta test this competition. Brought me back to the Farmville days, except on agentic steriods.
Build an AI agent to manage a virtual farm: harvest crops, care for animals, boost yields, expand your land, and trade on a dynamic market where prices react to supply and demand.
Kaggriculture, hosted by @Kaggle and @Google, is live!
💰 Total Prize Pool: $50,000
⏰ Entry Deadline: September 23, 2026
13
798
Walter Reade retweeted
I will be speaking at @AnalyticsVidhya DHS
Here's what I plan to cover:
✳️ What is a loop?
✳️ The anatomy of a loop
✳️ The four types of loops
✳️ Should this even be a loop?
✳️ A hands-on loop engineering session using a ML use case - fun one!
✳️ Use the loop, stay the engineer
Walter Reade retweeted
Great blog post from NVIDIA KGMON team @kagglingdieter @JFPuget summarizing practical lessons learned from the community in their Nemotron Model Reasoning Challenge hosted on Kaggle!
"1. Start with reasoning traces you can verify, not just more examples.
2. Treat token budget as part of the reasoning problem.
3. Use specialized solvers when a task has clear structure.
4. Validate against the failure modes you actually care about.
5. Make training choices that preserve reasoning behavior, not just leaderboard score."
developer.nvidia.com/blog/le…
Walter Reade retweeted
Kaggle (and the data we make openly accessible) provides an insightful glimpse into evolving ways of working with AI.
When incentives and evaluation criteria are clear, AI is an accelerant for novel discovery and human creativity (even as stylistic choices start to converge).
Here's an interesting study: An analysis of @kaggle code submissions by @gburtch showed AI coding agents are standardizing implementation details (such as using 42 as a random seed), but so far there is no evidence of homogenization of approaches and problem-solving strategies.
The Hitchhiker’s Guide to Monoculture
arxiv.org/pdf/2607.13077
Here's an interesting study: An analysis of @kaggle code submissions by @gburtch showed AI coding agents are standardizing implementation details (such as using 42 as a random seed), but so far there is no evidence of homogenization of approaches and problem-solving strategies.
The Hitchhiker’s Guide to Monoculture
arxiv.org/pdf/2607.13077
Walter Reade retweeted
As an aside, I love that action efficiency charts are becoming more mainstream
Intelligence is efficiency and we need new ways to represent information efficiency (not just cost!)
This is a step in the right direction
