At CAVE (Center for Automation, Virtualization & Embedded Systems), I designed and implemented a Digital Twin framework across Edge and Cloud layers for industrial control system monitoring. I modeled a Buck Converter power circuit using Runge-Kutta 4 numerical integration.
I built a local Edge controller with closed-loop PID control, achieving a 1.2 ms settling time and less than 4.5% overshoot under step-load changes. It remained resilient to simulated network drop-outs of up to 500 ms through local asynchronous buffering and queuing.
For the framework’s Cloud analytics, I developed a microservice to ingest telemetry, run anomaly detection, and generate dashboard visualizations. The project used FastAPI, Uvicorn, SQLite, and Jinja2.
My projects include AegisRAG, an enterprise knowledge assistant for document search and question answering, and a full-stack stock portfolio management system. I also built a fake news detection model using TF-IDF and Logistic Regression, with a Flask REST API for predictions.

