Zechen Zhang

I'm Zechen Zhang. My path has wandered through philosophy and physics, theoretical physics, evolutionary dynamics, and deep learning—each turn driven by the same question: how do complex systems learn and adapt?

That curiosity took me from studying the mathematics of evolution to the statistical mechanics of neural networks, working with Haim Sompolinsky at Harvard. More recently, my research shifted to how LLMs integrate new knowledge through fine-tuning—which led me to believe continual learning is the last piece of the AGI puzzle.

Along the way, I became deeply involved in AI safety—organizing Harvard's first AI alignment seminar series and remaining active in the Harvard AI Safety Team. I believe powerful AI is arriving faster than most realize. And if it's going to benefit humanity broadly, it can't remain a privilege gatekept by elite institutions with massive compute budgets and exclusive networks.

That's why I'm singularly focused on building Orchestra—AI scientific assistants that give everyone with a curious mind a Jarvis for science. It's the most urgent thing I can work on.