Shoumik Roychowdhury Portfolio

To evaluate whether combining maximum RSI-derived restriction scores (RSIrs-max) with deep learning (DL) models can enhance patient-level detection of csPCa compared to using PI-RADS or RSIrs-max alone.

MemZ uses AI to bring memories to life, transforming the way Alzheimer's patients interact with their past.Tools and Frameworks used: NeRF Studio, AWS Lambda, DynamoDB, EC2, React, Typescript, Groq, Hume.

Applied existing Mechanistic Interpretability techniques to retrieve interpretable visual and language representations of ascending VLA model reasoning informing final VLA model output [robot actuation commands]. Engineered novel interventions- inspired by existing Mechanistic Interpretability techniques - to make VLA models safer.

Developed an automated novel neural-network architecture and pipeline, aimed to quantitatively derive insights on router placement. Inspired from DensePose From Wifi(Geng, J. et.al)

Developed pipeline for simulating subsampled sparse k-space data, for fast MR reconstruction. Designed and developed a new GAN architecture, to simulate L-1 loss, via TensorFlow, with two groundbreaking innovations: introduction of a new perception loss function, and cross pollination of tensors. (ACM Cutler-Bell Winner)

An automated data-centric tool (AutoDC), similar to the purpose of AutoML, aims to speed up the dataset improvement processes. In our preliminary tests on 3 open source image classification datasets, AutoDC is estimated to reduce roughly 80% of the manual time for data improvement tasks, at the same time, improve the model accuracy by 10-15% with the fixed ML code.

Designed and engineered a series of CDCGANS simulating the rate of progression of Alzheimer's Disease and the atrophy of the brain over time. (Published IEEE IJCNN 2020)