Projects I've been cooking up

• 01

  Embedded TCN for Device-Level Arrhythmia Detection

  Implemented and trained temporal convolutional networks in PyTorch for cardiac time-series classification. Optimized for deployment on an Arduino Nano 33 BLE (Cortex-M4F) under hardware memory constraints — detecting AF, VT, Brady, and Pause arrhythmias at the edge.

  2nd Place — Stanford BASES Pitch Competition

  PyTorch Arduino Nano BLE Cortex-M4F Edge ML Time-Series

• 02

  Compute-Constrained Intelligence For Real Time Election Monitoring

  Shipping RNN/CNN, OCR, and NLP pipelines (Python, PyTorch, OpenCV) with Google Cloud Vision APIs to efficiently parse unstructured electoral reports and field notes. Developing time-series anomaly-detection models (ARIMA, statistical inference) to flag irregular voting patterns across multi-region real-time data (international elections). Pending integration into currently hard-coded Electronic Voting Machines (EVMs).

  Python PyTorch OpenCV Google Cloud Vision APIs

• 03

  International Financial Fraud Visualizations

  Applied regression analysis on IHDS survey data to predict wage dispersion across cultural and service-sector occupations; visualized results through income-distribution histograms and Lorenz curves. Collaborated with Dr. Lalita Du Perron and Dr. Eren Tasar on computerized simulations of financial fraud in India — high-dimensional transaction data, graph-based clustering, quantifying propagation patterns.

  Regression Graph Clustering Lorenz Curves IHDS Data

• 04

  Computer Architecture

  Something's cooking.

  coming soon

• 05

  Codesign Agents

  Something's cooking.

  coming soon