Hi, I’m Jai
(aka jkbhagatio)
Welcome to my personal site!
I’m interested in how stuff works, and how we can best learn how stuff works.
I try to take my work seriously, but not myself 😁
I’m a Founding Machine Learning Researcher at Metamorphic, building foundation models of the brain.
I like to learn, create, build, play, philosophize, science, research, program, solve, exercise, music, craft, meditate, ball, dance, teach, help, improve, and gen i.
01 / A bit of context
Creating and understanding intelligent systems
Generally, my research focuses on creating and understanding intelligent systems† — systems that make use of complex environments to achieve varied goals. Using animal nervous systems and artificial neural networks as models, I create and test theories that explain intelligent behavior as a function of processes that propagate information within these models‡.
Previously, I completed a Ph.D. in Computational & Systems Neuroscience at the Sainsbury Wellcome Centre (SWC) at University College London (UCL), an A.S.P. in Brain & Cognitive Sciences at Massachusetts Institute of Technology (MIT), and a B.A. in Neuroscience at Boston University (BU). I’ve also worked with Jack Lindsey (Anthropic) as a Bogue Fellow, Stefan Heimersheim (Apollo Research) as a MARS Scholar, and as a Data Scientist, Software Developer, and Technical Associate at UCL, the International Brain Lab, and MIT. In these roles, I’ve contributed to many renowned, collaborative research projects in the fields of Artificial Intelligence, Machine Learning, Neuroscience, and Neuroengineering.
A couple of notes on intelligence & models
† Roughly, I think of intelligence as the general and practical acquisition, recall, and application of knowledge and skills in familiar and novel environments in pursuit of goals.
‡ In animals, we examine electrical activity that propagates sensory information within the nervous system and yields behavior. In artificial neural networks, we examine the numerical values of the activations of and connections between individual units that make up the network; we examine this activity purely in software, though I think it should also be examined in conjunction with the underlying electrical activity that takes place in computer hardware.
02 / Active research areas
01 / Models
How can we train models to learn structure across neural activity, behavior, and the world?
03 / Additional research interests
01 / Interfaces
How might we exchange information between brains and machines?