Google’s AI Solves 56-Year-Old Math Problem, Codex Writes Code and Manus Makes Art That "Reads" Your Mind
Google builds a model that invents algorithms, OpenAI debuts a GitHub-savvy coding agent and Manus tackles multi-step creative tasks.
Fellow Data Tinkerers!
It’s time for this week’s round-up on all things data and AI. Before that, just sharing that you can have access to 100+ cheat sheets like below, if you share Data Tinkerer with just 2 other people
So if you enjoy reading Data Tinkerer, share with friends and earn referral rewards!
Now, with that out of the way, let’s get to this week’s news round up on all things data and AI
The Buzz 🐝
Big news out of Google last week was AlphaEvolve. So what’s all the hoopla about? It goes beyond basic code generation by actually evolving algorithms using LLMs plus evolutionary search. So in short, it’s an algorithm inventor rather than just a code generator.
It has already increased data center efficiency for Google by 0.7% and improved solutions for 20% of 50+ open math problems, including breaking a 56-year-old record in matrix multiplication. You can read more about it here
AlphaEvolve used 15 complex mutations across multiple components to find faster matrix multiplication algorithms. (Source: Google) Another major news last week was OpenAI’s Codex. It’s AI coding agent (yes, another agent) built on the o3 model that writes features, fixes bugs, answers code questions and run tests. While some early users have had positive feedback, some others are not so sure about its differentiation and long-term impact:
(Source: LinkedIn) If you wanna get started with Codex, check the video below by the OpenAI development team.
And the last piece of news for this week is about Manus AI. Manus is now open to everyone with no waitlist, free daily tasks and a 1,000-credit bonus. Plus, They rolled out image generation which goes beyond just creating an image. It tries to understand your intent, plan solutions and use other tools to get the job done. You can check the demo video below:
Data Science & AI
How Reddit Scans 1M+ Images a Day to Flag NSFW Content
How Reddit Scans 1M+ Images a Day to Flag NSFW Content Using Deep Learning
·
Reddit needed to flag NSFW images the second they were uploaded. They built a deep learning system that does exactly that; fast, scalable and battle-tested in prod. Here’s how it works.
How On-Device AI Models Find Your Best Tinder Profile Photos
Learn how Tinder uses on-device AI to find, verify and recommend your best profile photos to boost matches while keeping your data private- provides an inside look at building a career in AI, sharing how perseverance, openness and steady writing over eight years led from outsider status to key roles at HuggingFace and Ai2.
Data Engineering
Behind the Scenes: Building a Robust Ads Event Processing Pipeline
Learn how Netflix built a centralized ads event pipeline that streamlines tracking, reporting and billing - enabling faster launches and reliable ad measurement.
How Canva Rebuilt Its Data Pipelines for Billions of Events per Month
Canva had to track billions of events to pay creators fairly and their old system couldn’t keep up. Curious how they rebuilt it? This article is for you
If you're learning Kafka, this article is for you
provides a great baseline for learning Kafka, breaking down its architecture, producer and consumer mechanics.
Data Analysis and Visualisation
The Most Popular Web Browsers
Interesting visualisation of the most popular web browsers
When the Metric Becomes the Monster
Learn how Goodhart’s Law quietly wrecks your metrics. When the target becomes the game, teams start optimizing for the number, not the outcome. Here’s how to spot it and what to do instead.
"o3, show me a photo of the most stereotypical X and LinkedIn feeds as seen on a mobile device. Really lean into it."
If you enjoyed reading this week’s round up, please give this post a like and share it with others
