Why “CrossCurrents?”

This site was an outgrowth of my personal portfolio, designed after I reached a critical threshold on my research journey. I’m inspired by the documented positive effects of multimodality on LLMs. What is it about processing terabytes of image data that makes LLMs better at coding? What are the big, broad ideas about the structure of the world that these algorithms are drawing out and successfully applying? These are the world’s “general principles,” and one of the most difficult parts of human-conducted research today is prioritizing the search for these principles over the pursuit of incremental improvement.

In my view, the most natural way to discover these principles is to do exactly what the LLMs are doing: learn from the massively abundant human knowledge that exists outside our individual specializations. Going where we’re not supposed to, learning what we don’t have to, and making the mistakes we’ve been afraid to, can be some of the most effective ways of furthering our understanding of the world. More than learning specific facts, our aim must be to learn the cognitive toolkits and perceptual lenses which led to the discovery of those facts. That is to say, it is not enough to simply learn biology; we must become biologists, become engineers, become economists, hackers, and businesspeople. The accumulation of these cognitive toolkits and perceptual lenses—humanity’s general principles—is far more powerful than the sum of its parts, and enables us to recognize, approach, and solve problems in ways that were previously impossible.

Hence, CrossCurrents: a project intended to further the exchange of information across disciplines. I will butt my head where it doesn’t belong, be confidently wrong because of my preconceptions, and publicize my mistakes so that others might learn. I will stretch my cognitive toolkit past its point of failure, and from its broken pieces construct something more robust. My aim is to learn everything I can about the world around me, synthesize those lessons into sharp (and occasionally provocative) opinions, and test those opinions against those of more experienced specialists.

I expect and welcome criticism, and have enabled comments on all of my posts. I enforce civility, professionalism, and meaningful contribution to the discourse via a minimum character count. Other than that, I welcome any and all contributions from researchers, industry professionals, or anybody else, regardless of background or field of study. My goal is to be proven wrong, so that I can correct my assumptions.

Finally, although I believe discussion and debate is necessary to build a better understanding of the world, I do not believe it is sufficient. For that reason, I will also post regular updates on the real-world technical projects I’m working on. This could be research, open-source contributions, or small side-projects that I find to be interesting or useful demonstrations of important general principles. Though because this project is an outgrowth of my old portfolio, I would caution you against scrolling too far down, unless you want to see my work become much less defensible and much more embarrassing. I have refrained from deleting my old content, as one of my goals for CrossCurrents is to publicize my mistakes. Perhaps someone might find it useful to read a very, very bad article on how I set up my original website (thankfully no such article is needed for the current one; Claude did all the work).

Who am I?

Unfortunately, knowing where I’m coming from is essential for my readers to evaluate my ideas effectively. I suppose I couldn’t avoid the question forever. I’m Nikhil, an undergraduate student majoring in computer science and biology at the University of Texas at Austin. I’ve been engaged in research almost continuously since high school in fields ranging from geosciences to computational biology to aerospace engineering, but I would say my primary unifying interest is machine learning. You can find my resumé here to get a more detailed overview of my background and experience. If you like what I do, hate what I do, or have any other opinion in between, please get in touch!

Connect with me on LinkedIn here: https://www.linkedin.com/in/nikhil-kalidasu/. If you’d like to see some of my contributions to the open-source community, check out my GitHub. You can email me at srikarnikhil.kalidasu@gmail.com, and I will do my very best to get back to you.