Developed an AI-powered web application to translate Pakistan Sign Language (PSL) into text and speech, improving accessibility for differently-abled individuals. Built using the MERN stack and integrated TensorFlow Object Detection for real-time sign recognition. Fine-tuned AI and Computer Vision models, including YOLOv5, Faster R-CNN, EfficientDet-D0, MobileNet-SSD, and LSTM. Created a chatbot using LangChain, fine-tuned on custom FAQs data and integrated with OpenAI APIs, to assist users with adaptive, intelligent responses. Implemented Redux for state management and Mongoose for efficient data handling. Led team collaboration, managing task distribution, timelines, and technical communication. Engineered core backend features including schema design, API integration, authentication systems, and robust testing workflows.

Developed a Multi-Agent System using LangGraph, Flask APIs & scikit-learn to detect and mitigate algorithmic and data biases in healthcare AI systems. The system ensures ethical compliance through fairness metrics, real-time analysis, and a roadmap for transparent and inclusive AI development.

Developed a clinical decision support system that classifies medical queries and routes them to specialized tools for data extraction. The retrieved data is then processed by a final language model to generate accurate, clinically-backed responses.

Developed and evaluated multiple machine learning models (classification + regression) on a dataset of 500,000+ patients to predict liver disease progression. The project focused on early diagnosis by leveraging key health indicators and applying robust data preprocessing and ensemble modeling techniques.

Enhancements in the local Nexus Project by developing real-time PySpark ETL pipelines and FastAPI-based microservices for efficient data flow across multiple hubs. The project aimed to streamline data processing and improve scalability within healthcare systems.

Analyzed over 600k patient records across U.S. states to proactively identify chronic disease risk hotspots using advanced data preprocessing and statistical analysis. The project provided key public health insights by mapping disease trends and supporting preventive care strategies.

A fully automated AI-powered voice receptionist for clinics and hospitals that handles inbound patient calls end-to-end; collecting essential information, and seamlessly scheduling appointments with the relevant doctors. The system supports multiple languages, making it accessible to a diverse patient base. It is equipped with intelligent tools that allow it to verify and cross-check patient data directly from the database in real time, ensuring accuracy throughout the conversation. The system delivers a smooth, human-like experience. A dedicated Admin Dashboard provides clinic staff full visibility and control over call recordings, patient data, and appointment management; all in one place.

A Multi-Agent System that automatically analyzes call center agent conversations against key KPI metrics. The system evaluates each call, identifies performance gaps, and autonomously generates personalized coaching and training recommendations for each agent; helping them understand their weaknesses and continuously improve without requiring manual supervisor intervention.