I am a Computer Science undergraduate passionate about applying Artificial Intelligence to scientific and real-world problems.
My primary interests lie at the intersection of:
- Computer Vision
- Healthcare AI
- Physiological Signal Analysis
- Geospatial Machine Learning
- Deep Learning for Scientific Applications
I enjoy building machine learning systems, conducting applied research, and exploring how data-driven methods can contribute to healthcare, environmental sciences, and intelligent perception systems.
research_interests = {
"Healthcare AI": [
"Physiological Signal Analysis",
"Medical AI",
"Multimodal Learning"
],
"Computer Vision": [
"Image Classification",
"Scene Understanding",
"Representation Learning"
],
"Scientific AI": [
"Geospatial Machine Learning",
"Remote Sensing",
"Environmental Modeling"
],
"Emerging Interests": [
"EEG Analysis",
"Neuroimaging",
"Brain Signal Processing"
]
}Department of Earth & Environmental Sciences
Worked on machine learning approaches for environmental risk assessment using heterogeneous geospatial datasets.
- Developed hybrid ML pipelines using Random Forest, XGBoost, and SVM
- Performed terrain-aware spatial feature engineering
- Designed spatial cross-validation strategies
- Worked with large-scale environmental datasets
- Utilized GeoPandas, GDAL, ArcGIS Pro, QGIS, and Scikit-Learn
- Reduced critical-region misclassification below 5%
Python GeoPandas GDAL QGIS ArcGIS Pro Scikit-Learn
- Presented at ISCMCTR 2025
- Accepted for publication in MITS Journal 2026
Comparison of:
- Logistic Regression
- SVM
- Random Forest
- LSTM
- BERT
- RoBERTa
for financial sentiment analysis tasks.
Current ongoing research exploring:
- Multimodal physiological signal analysis
- Deep learning for healthcare
- Temporal modeling of biosignals
- Explainable AI
- Early disease detection systems
- EEG-based biomarker discovery
Developed a multimodal biosignal-based sleep apnea detection system using:
- Airflow signals
- Thoracic movement signals
- SpO₂ measurements
- Implemented 1D-CNN models
- Developed Conv-LSTM architectures
- Subject-wise LOPO validation
- Achieved 85% accuracy
- Achieved 88% sensitivity
- Extended toward sleep stage classification
Python PyTorch TensorFlow Deep Learning
Developed a research-oriented computer vision pipeline for robust scene-independent change detection and foreground motion segmentation. The project focuses on identifying meaningful scene changes while remaining resilient to variations in illumination, background dynamics, camera viewpoints, and environmental conditions.
The objective was to learn generalized visual representations capable of distinguishing true scene changes from nuisance factors, enabling reliable deployment across previously unseen environments.
- Developed a deep learning pipeline for robust foreground change detection
- Designed temporal feature extraction and attention-based modeling
- Implemented ResNet18 encoder-decoder architecture
- Incorporated edge-aware supervision for sharper change localization
- Explored self-supervised learning strategies to reduce annotation dependency
Python PyTorch OpenCV Deep Learning Computer Vision ResNet18 Temporal Attention Self-Supervised Learning
End-to-end Earth Observation pipeline for satellite imagery analysis.
- Satellite image processing
- Spatial feature engineering
- Coordinate transformation pipelines
- ResNet-based classification
- Achieved 87% classification accuracy
Python Remote Sensing Deep Learning Computer Vision
Developed an autonomous navigation agent capable of learning directly from visual observations.
- Deep Q Networks (DQN)
- Experience Replay
- Target Networks
- End-to-End Learning
- Reinforcement Learning
- Python
- C
- C++
- SQL
- MATLAB
- PyTorch
- TensorFlow
- Keras
- Scikit-Learn
- OpenCV
- Time-Series Analysis
- Feature Engineering
- Physiological Signal Modeling
- Frequency Domain Analysis
- Digital Signal Processing
- Multimodal Data Fusion
- GeoPandas
- GDAL
- QGIS
- ArcGIS Pro
- Raster Processing
- Spatial Analytics
- Git
- Docker
- Jupyter Notebook
- Linux
- Google Colab
- LaTeX
- Oracle Cloud Infrastructure Generative AI Professional
- Machine Learning with TensorFlow (Infosys Springboard)
- Python Certification (IIT Bombay Spoken Tutorial)
- Summer Research Intern — IISER Mohali
- Research Paper Presented at ISCMCTR 2025
- Publication Accepted — MITS Journal 2026
- Finalist — AI Hackathon, IIT Kharagpur
- Finalist — NLP Hackathon, IIT Kharagpur
- Finalist — Finance AI Hackathon, IIT Kanpur
currently_working_on = {
"Research": [
"Healthcare AI",
"Physiological Signal Analysis",
"Computer Vision"
],
"Learning": [
"Neuroimaging Fundamentals",
"EEG Analysis",
"Advanced Deep Learning"
],
"Building": [
"Research Projects",
"Open Source Contributions",
"Applied AI Systems"
]
}
