Academic Portfolio

This paper introduces a new  design philosophy that counters the "deskilling" caused by friction-free automation by prioritizing human capacity building over immediate task completion. Grounded in educational psychology concepts like Vygotsky’s Zone of Proximal Development and Bjork’s Desirable Difficulties, the framework models AI as a pedagogical agent that supports rather than replaces human cognition. .

This paper proposes a control-theoretic framework that treats HMI as a partnership where both agents actively minimize mutual confusion (variational free energy) rather than simply exchanging commands.
It introduces the Co-Adaptive Intent Loop (CIL), which uses recursive Bayesian estimation to proactively infer latent user intent while simultaneously signaling the machine's uncertainty back to the user through "Fluid Legibility" cues.

This paper proposes a framework that moves beyond the traditional "user-tool" metaphor to treat the human and AI as a single, unified cognitive system rather than a master-slave dyad.  

It utilizes principles of distributed cognition to seamlessly share processes like memory and attention between the user and machine, minimizing cognitive friction through bi-directional adaptation and shared intentionality.  

Developed a multimodal attention system that resolves the "Grounding Problem" by correlating human pointing gestures with semantic object labels in real-time.

Implemented a hybrid pipeline using MediaPipe (Skeleton Tracking) and TensorRT SSD (Object Recognition) on an embedded edge device.

Engineered a vector-projection algorithm to map 2D user intent onto physical world coordinates, simulating joint attention mechanics.

A phygital interface that turns painting gestures into live, generative Raag Yaman music for creative augmentation and flow. A brush is tracked via webcam in HSV space using OpenCV and NumPy, with its position and motion mapped in real time to pitch (quantized Yaman scale), stereo panning, and dynamics through SCAMP + FluidSynth. The result is a tight sensorimotor loop where invisible brushstrokes become evolving raga-based sound, fusing bodily motion, visual feedback, and musical response into a single flow-inducing instrument.

Using Organic User Interfaces (OUI) this project transforms a botanical environment into a responsive, multimodal instrument. By instrumenting 12 living plants with discrete electrodes via a Bare Conductive Touch Board, the installation utilizes the natural capacitive properties of flora to replace traditional rigid controllers. Tactile interactions with the plants trigger real-time MIDI data streams, which are routed to FL Studio (via Midify) to synthesize generative Indian Classical music compositions, effectively sonifying the biological contact. 

We are developing a multimodal cognitive rehabilitation solution for patients with stroke, TBI, and neurodegenerative disorders, integrating non-invasive BCI, VR, and Generative AI-powered gamified software.

We are developing a completely enclosed Brain-Computer Interface and VR headset system designed to support children with autism. This platform utilizes a multimodal approach by integrating fNIRS, eye tracking, motor function data, and cognitive function assessments, all based on five distinct classes of datasets

We are developing an invasive Brain-Spinal Interface in collaboration with mPragati, IIT Delhi, and AIIMS spinal cord surgeons to capture motor intentions from the brain and use them to stimulate the spinal cord. Also working on a review paper on the same.

We have developed a cognitive augmentation tool that uses a Sony Spatial Reality Display & Unity to create a high-fidelity 3D 'memory palace' that enhances memory by serving as an external platform for the 'method of loci' mnemonic technique, thereby reducing the cognitive load required to mentally visualize space.

This study systematically analyzes over 200 first-person testimonies of encounters with the sage Sri Ramana Maharshi to investigate reports of profound, presence-induced changes in cognition and emotion. Using a mixed-methods approach, the research identifies six consistent themes from these encounters, including the involuntary cessation of thought ("The Primacy of Silence"), an instantaneous shift to peace, and a dissolution of egoic identity. 

This research investigates the idea that mathematics is not purely abstract, but is fundamentally rooted in human cognition and metaphorical connections to everyday concepts. The study replicates previous work using computational models to analyze the semantic structure of mathematical words, confirming some findings but also revealing the limitations of current models like GloVe.