Sai Rithvik Kanakamedala
Data Scientist | Advanced Analytics | ML Engineering
MS Data Science from Columbia University (3.99 GPA) • Currently driving data-driven decisions at Novo Nordisk through ML-based propensity models and patient analytics • Award-winning researcher with published papers and hackathon victories
Technical Expertise & Impact
Transforming complex data challenges into actionable insights through advanced analytics and machine learning
Machine Learning & AI
- Python & PyTorch
- AutoML Implementation
- Propensity Modeling
- Deep Learning Algorithms
Data Engineering
- PySpark & Snowflake
- Dataiku Platform
- Google Cloud Platform
- ETL Pipeline Design
Analytics & Visualization
- Statistical Inference
- Tableau & Plotly
- SHAP Interpretability
- Advanced Analytics
Healthcare Analytics
- Patient Analytics
- HCP Targeting
- Market Access Analysis
- Clinical Data Insights
Efficiency improvement through AutoML implementation
Sales lift from predictive modeling solutions
GPA MS Data Science academic excellence
Researcher with conference presentations
Professional Experience
Novo Nordisk
- Led ML-based propensity models for HCP targeting
- Presented at PMSA Fall Symposium
- Automated end-to-end AutoML process
Novo Nordisk
- Developed classification model for obesity medication prescribers
- Demonstrated 7% sales lift through test-control study
IIT Madras
- Achieved 72% accuracy on CIFAR-10 with CNN implementation
- Enhanced test accuracy by 3% through distance minimization
Data Science Projects
Demonstrating practical application of skills through detailed project showcases with technical depth
Adverse Drug Reaction Analysis
Award-winning project utilizing advanced NLP and ML techniques for pharmaceutical safety analysis.
- Award-winning project at Columbia Data Science Hackathon
- Best among 25 teams
- Advanced NLP and ML techniques for pharmaceutical safety
Loan Default Prediction System
High-performance binary classification model achieving exceptional accuracy in loan default prediction.
- Binary classification model with 88.9% accuracy
- KPCA dimensionality reduction
- SMOTE for class imbalance
- AUROC 0.9 performance
Medical Image Segmentation
Innovative Double U-Net architecture with significant parameter reduction and published research results.
- Double U-Net architecture with 50% parameter reduction
- Custom activation function
- 0.7135 dice similarity coefficient
- Published at ICVISP conference
Ready to Collaborate?
Currently open to consulting opportunities and research collaborations in healthcare analytics and machine learning. Let's discuss how data science can drive your next breakthrough.