Alex Smola

Tomorrow I am giving one of the invited talks in the Applied Data Science track at KDD 2026 in Jeju, Korea.

A bank’s voice agent has verified the caller and is reading out which accounts have a pending overdraft decision.

Last post was the story of dragging MXNet out of the Apache Attic, and the method I used to do it: run the tests, treat every failure as a clue, write a regression test for each bug, and let the agents do the legwork.

Apache projects don’t really die.

Next week I am teaching a tutorial on efficient LLM inference at the Machine Learning Summer School 2026 in NYC, hosted this year at Columbia University.

Yesterday’s post closed with a conundrum: all three estimators assumed that computing is cheap, yet a dense Gaussian projection to dimensions costs multiplications per vector.

A surprising number of posts on this blog secretly run on the same primitive.

When was the last time you enjoyed calling your cable company?

Years ago I wrote about correcting covariate shift by reweighting your data.

A few days ago I wrote about conversational proactivity: a model noticing what you disclosed but never asked about, and turning it into something useful.

A user has spent ten minutes preparing for a small art installation.

One of my favorite interview questions for network engineers (no longer, now that I’m about to spoil it): you have 512 GPUs across 64 servers, eight GPUs per box.

The natural reaction to the Pokémon theorem is to escape into representation learning.

If you have been near algorithmic fairness for the past decade, you already know the punchline: you cannot have all the things you want.

Every time a new model comes out, somebody runs it on MMLU (57 subjects), MTEB (56 tasks), HELM, the Open LLM Leaderboard, AlpacaEval, LiveBench, BigCodeBench, WildBench, Arena-Hard, MT-Bench, and a dozen others.

I’m happy to announce our new book project - Dive into Deep Learning.

Dear Friends, As some of you may have already heard, I’m leaving CMU to join Amazon, effective July 1, 2016.

One of the key features of a parameter server is that it, well, serves parameters.

We’ve been busy building the next generation of a Parameter Server and it’s finally ready.

Disks are slow and RAM is fast.

The Weisfeiler-Leman algorithm and estimation on graphs Imagine you have two graphs and and you’d like to check how similar they are.

In defense of keeping data private This is going to be contentious.

The videos from MLSS 2011 in Purdue are now available online.

Many algorithms require random number generators to work.

So you read a book on Reproducing Kernel Hilbert Spaces and you’d like to try out this kernel thing.

We’re organizing a workshop at NIPS 2011.

Here are the slides [Keynote, PDF] for a basic course on Graphical Models for the Internet that I’m giving at MLSS 2011 in Purdue that Vishy Vishwanathan is organizing.

Here’s a simple synchronization paradigm between many computers that scales with the number of machines involved and which essentially keeps cost at per machine.

The code to our LDA implementation on Hadoop is released on Github under the Mozilla Public License.

Bloom filters are one of the really ingenious and simple building blocks for randomized data structures.