Surabhi Pandey
Healthcare AI | Biomedical Signals | Computer Vision
Building interpretable and clinically reliable AI systems.
Healthcare AI | Biomedical Signals | Computer Vision
Building interpretable and clinically reliable AI systems.
I’m Surabhi Pandey — a Machine Learning student at IIT Guwahati focused on research-driven AI systems in healthcare, biomedical sensing, and computer vision. My work centers on building models that are not only accurate, but clinically meaningful and technically rigorous.
I am particularly interested in medical imaging, prosthetic and wearable sensor data (IoT-based systems), and interpretable deep learning. I enjoy studying how models behave under real-world constraints such as patient variability, noisy signals, and domain shifts.
Indian Institute of Technology, Guwahati
2023 – 2027
Machine Learning Intern – Hybionics Pvt. Ltd.
February 2026 – Present
Working on prosthetic IMU sensor signal modeling for motion analysis and fall-risk detection.
Includes preprocessing, windowing, feature extraction, and robustness evaluation on real-world biomedical time-series data.
Machine Learning Intern – BioScanAI Pvt. Ltd.
October 2025 – January 2026
Developed and evaluated machine learning models for retinal imaging–based disease classification.
Performed data preprocessing, feature engineering, model validation, and performance benchmarking to improve robustness and predictive performance.
Trained a ResNet50 model for brain tumor classification (94% test accuracy) using patient-level splits. Integrated Grad-CAM and segmentation-guided counterfactual MRI generation to quantify causal dependence using Delta-Drop and Delta-Focus metrics.
Grad-CAM RepositoryFull-stack healthcare application with ML-based disease prediction. Built using Flask API, MySQL backend, and HTML/CSS frontend for managing patient records and symptom-based predictions.
View on GitHubWeb-based statistical testing system allowing CSV upload and automated hypothesis testing with structured analytical report generation. Supports classical statistical workflows with interpretive summaries.
View on GitHubGemini API-powered conversational system deployed on HuggingFace. Includes crisis keyword detection and intelligent response generation for supportive mental health interaction.
View on GitHubFor more projects, visit my GitHub.
You can explore my full CV to see research projects, technical skills, and certifications.
Download My CV
