Published:
At HardMode 2026 at MIT Media Lab, our team built DriftPak — an AI-powered backpack that can see, hear, speak, and know where it is. Here’s the full story of how we went from a Pi and a pile of ESP32s to a sentient bag.
Published:
The panic a lot of knowledge workers are feeling is not about whether AI will take their jobs. It is about whether the years they spent learning to do hard things manually were worth it. I think the answer is: it does not matter anymore. There is a door right there.
Published:
At a Google DeepMind x Cactus Compute hackathon, I used Claude Code to scrape every competitor’s fork, benchmark all 45 modified solutions, and synthesize a top-scoring submission from the best of them. This is a vulnerability disclosure.
Published:
Team MindMeld placed 1st at BrainStorm 2026 by decoding 1024-channel brain recordings in real-time on edge hardware. Here’s the full technical breakdown of our winning approach.
Published:
If you use AI chat tools regularly, you’ve run into this. You have two or three conversations going in parallel, maybe originally about different things, and at some point you realize they’ve converged. You want one chat to be able to see what you discussed in another. But there’s no way to do that. The conversations are siloed. You can copy-paste context across windows, but that’s tedious and lossy. You can try to summarize one chat and feed it to another, but you lose nuance.
Published:
Claude Code sessions silently fragment when you cd between directories or exit and resume. I built session-merge to detect and fix split sessions — merging them back into one continuous conversation.
Published:
I’ve been using Claude Code heavily for the past few months — for research projects, web apps, scripts, even writing. And I completely get it now when the creators say they’re using Claude Code to build Claude Code. Once you start working with it seriously, the meta-loop becomes impossible to ignore.
Published:
MIT Deep Learning final project exploring whether pre-trained unimodal encoders can be aligned into a shared multimodal latent space using lightweight adapters.