Recurrent neural networks (RNNs) for the inference of neural state transitions
For my PhD in Rajan Lab, I worked on building computational tools using recurrent neural networks to help in the inference of neural state transitions and circuit mechanisms which could be predictive of behavior, and could further be used in various clinical and brain machine interface applications. I also applied these tools to human electrophysiological data from deep brain stimulation patients in Mayberg Lab. A couple of relevant links to my PhD research [1][2].

Development of algorithms for closed-loop neurostimulation in Parkinson's disease
I worked as a research engineer at Stanford University School of Medicine in Dr. Helen Bronte-Stewart's lab. The lab uses deep brain stimulation (DBS) to control various motor symptoms of Parkinson's disease. We recorded local field potentials from the subthalamic nucleus of the basal ganglia of the patients. I worked on finding neural biomarkers related to various motor symptoms (tremor/bradykinesia/freezing of gait) using signal processing. In addition, I developed and tested algorithms for closed-loop neurostimulation to control motor symptoms. 

Recurrent neural network models of voluntary motor control
For my MS research, I worked in Dr. Ali Minai's lab at University of Cincinnati. I developed recurrent neural network models motivated by the need to understand volitional motor control. We showed reliable storage and recall of aperiodic spatiotemporal activity patterns in these networks, which is probably one of the ways how neural systems encode information for motor control.

Development of automated visual reinforcement audiometry (VRA)
I worked in Dr. David Moore's lab at Cincinnati Children's Hospital Medical Center to automate visual reinforcement audiometry (VRA) using corneal reflection eye tracking for the detection of listening difficulty in children.