At Indian Institute of Technology, Indore, I architected and deployed a TabTransformer early-warning model for ventilator-associated pneumonia detection in the DT-CDSS initiative. It delivered a 5× precision-recall gain using leakage-safe temporal features and patient-level cross-validation.
I externally validated the model on MIMIC-IV, with reported high-80% sensitivity and specificity and cross-hospital generalization. The model was presented to clinicians and academic collaborators, and a decision-support system was deployed at AIIMS ICU.
As a Deep Learning Research Intern at Indian Institute of Technology, Indore, I built an ensemble CNN-Transformer framework for plant disease image segmentation and classification. I also automated its evaluation pipeline and deployed it to remote servers.
At HIXAA Technologies Pvt. Ltd., Nagpur, I developed a carpooling mobile application with real-time ride matching using Flutter and Firebase. My projects include ASIS, a desktop AI assistant with isolated Docker code execution, and WritoArms, a Flutter platform integrating user writing with AI-generated artwork.
