Viewing profile — ramaseshanms

People will still own the ideas. The way they are shared is changing yet again. Just like a kindle carried a Book to a user, and LLM with RAG will do as well. Printing books wouldn…

Local models Inference never really took with non tech people. Everyday people who dont know the difference between autocorrect and GPTs. But thanks to recent hardware launches fro…

Been waiting for this for so long. Huge respect to the team for pulling this off.

Its possible that Andrej Karpathy could have been hired for scaling his vision on the auto-research repo. (His version of "AI that builds itself")

Indeed, the simplicity and structure behind RSS perfectly fits agentic AI. Web 4.0 (or Web 5.0) would definitely use a protocol whose principles might be similar to RSS.

Standardizing is not quite simple. Each rocket is different from each other. There are many reasons why launchpads are intricately designed as per specific rockets. examples like T…

Yes, I meant the same. Basically, I called for expansion of existing spaceports and a handful of new ones. They will be closer to the coast as always with the launch paths facing t…

In order for rocket launches to reach 1000+ per year in a decade, we need more innovations and business models in launch pads. 18 months to restore a launchpad is going to hamper p…

Genuinely impressed with the article. We need more such articles that re-invigorate our engineering curiosity. Kudos!

How not to flex: "MAI-Code-1-Flash outperforms Claude Haiku 4.5"

The two-tower Mixture-of-Transformers design (autoregressive reasoner feeding a diffusion generator) is an interesting architectural bet.

The interesting design question here is whether unifying GUI and CLI operation in a single agent loop actually improves performance or just makes the benchmark story cleaner.

[flagged]