Siddartha Bandi
Building intelligent systems that turn raw data into real-world impact. Passionate about deep learning, satellite analytics, and scalable ML pipelines.
About Me
I'm a Data Scientist and AI Engineer currently pursuing my MS in Data Science at the University of North Dakota. My work lives at the intersection of machine learning, computer vision, and large-scale data engineering.
I've engineered automated Python pipelines that process decades of satellite imagery, built deep learning systems for medical image analysis, and developed risk models that directly reduced compliance overhead at a leading financial institution. I love turning messy, high-dimensional data into clean, actionable intelligence.
Outside of research and engineering, I'm passionate about open-source, reproducible science, and making AI tools accessible to domain experts across environmental, medical, and financial fields.
Technical Skills
Languages
ML / DL Frameworks
Cloud & DevOps
Data & Viz
Full Technology Stack
Featured Projects
Satellite Imagery Analysis Pipeline
35 Years of Earth History, Automated
Engineered an end-to-end automated Python pipeline to ingest, preprocess, and analyze over three decades of historical Landsat satellite imagery (1988–2023). Computed NDVI and EVI spectral indices at scale to track long-term vegetation dynamics and land-use change — transforming a months-long manual process into a reproducible, parallelized workflow.
Key Outcomes
- ▹Automated processing of 35+ years of Landsat scenes
- ▹Parallelized NDVI/EVI computation across temporal stacks
- ▹Produced publication-quality change-detection maps
Medical Image Analysis with Deep Learning
ResNet · YOLO · U-Net Across 3 Clinical Domains
Built a modular deep learning pipeline for multi-task medical image analysis across stroke, cardiac, and liver datasets. Deployed ResNet-50 for classification, YOLOv8 for lesion detection, and U-Net for precise organ/pathology segmentation — validating each architecture's performance on clinical benchmarks with rigorous cross-validation and metric reporting.
Key Outcomes
- ▹Implemented ResNet, YOLO & U-Net in a unified codebase
- ▹Achieved strong Dice scores on stroke/cardiac/liver segmentation
- ▹End-to-end pipeline from DICOM ingestion to metric visualization
Mashreq Bank Transaction Risk Models
35% Reduction in Manual Compliance Workload
Developed and deployed predictive risk models for Mashreq Bank using Python and SQL to flag high-risk financial transactions before compliance review. Combined feature engineering on behavioral transaction patterns with gradient-boosted classifiers, reducing the queue of manual reviews by 35% and improving the precision of risk flagging across product lines.
Key Outcomes
- ▹35% reduction in manual compliance review volume
- ▹Built SQL + Python feature store from raw transaction logs
- ▹Gradient-boosted models with real-time scoring integration
Get In Touch
I'm actively looking for Data Science, AI/ML Engineering, and Data Engineering roles. Whether you have a position in mind or just want to connect, I'd love to hear from you.
Internships, full-time roles, research collaborations — all welcome.