At Thapar Institute of Engineering and Technology, I engineered an asynchronous pipeline for physiological signal data and extracted RR intervals and respiratory features from ECG-derived respiration signals. My SVM, Random Forest, and LSTM ensemble reached 95% test accuracy for apnea event prediction.
I also built reusable processing modules for the PTB-XL and MIT-BIH datasets, eliminating more than 20 hours of manual preprocessing each week.
On IntelliFlow, I developed an AI workflow platform that turns client requests into structured projects, with task breakdown, employee matching, and approval workflows. I built its React and TypeScript frontend, Node.js API, and Python FastAPI AI microservice using LangGraph and Groq LLMs.
For my keystroke authentication project, I developed a behavioral biometric system using typing patterns and benchmarked several machine learning models, achieving 94% accuracy. At EDC, TIET, I coordinated E-Summit event operations and led business modeling and pitching workshops.
