This is what I learned after being a mentor in Stanford Code in Place for 3 years 👇

This 10-minute video can help you to work with Google Antigravity and deploy applications in minutes youtube.com/watch?v=ooHyVr…

The best resource to understand LLMs architecture!

Thinking in models in terms in intention and how you will use it helps to understand if it meets your standards

Code in place is back! Learn Python with Live interactive sessions following the CS106A Stanford curriculum. Apply as a student or a mentor codeinplace.stanford.edu/public/join/ci…

Remember 2020, when doctors and nurses were the world’s heroes? It was cool while it lasted.

Now we can play safely!!

I incorporate into my workflow for creating an application by starting with drawing the frontend, which helps to visualize what it looks like before putting everything into the code.

The vision behind Open Source LLMs. This is one of the most fun papers I have found. It shows practical examples and how it was built. youtube.com/watch?v=zk0Xpf…

2026 content creation strategy: don't look like a bot

This is amazing!! You can get intuition before jumping into the code. For me, building intuition is a key predictor of learning and application.

If coding is death, does that mean this Kaggle competition will be won by a PM?

AI By Hand is beyond a learning tool Prof. Tom Yeh follows a six-level approach 1. Black box 2. Components 3. Tensors 4. Math 5. Excel 6. Coding This allows incremental understanding and core foundational knowledge

This is one of the most impressive research I saw.

If you are building applications with Gemini models, a cool hack is to point at the latest version rather than change with every new release. Instead of doing gemini-2.5-flash do: gemini-flash-latest

AI has dramatically accelerated the "Model Implementation" step, which is huge progress! Yet the real value (and the real bottleneck) in production ML systems still lives in everything around that box: reliable data pipelines, thoughtful feature design, robust serving

Bard was fun!

Why people randomly download skills instead of copy paste the raw instructions?

The advice of not learning to code because you won’t code by hand is the equivalent to not learning math because calculators can do it for you. The difference is that calculators are 100% correct

What's the most used AI tool in Japan?