Abhijith Kumar Email & Phone Number

United States

St Louis, Missouri, United States

• Predictive Modeling for Healthcare Resource AllocationProblem Statement: Optimize healthcare resource allocation to improve efficiency and reduce costs.Responsibilities:
• Ingested healthcare data (e.g., claims data, patient demographics) into Google Cloud Storage using CloudDataflow.
• Processed and cleaned the data using Dataproc and Spark to prepare it for modeling.
• Developed a predictive modeling framework to forecast healthcare resource needs.
• Integrated the predictive model into resource allocation planning processes.Natural Language Processing for Clinical Note AnalysisProblem Statement: Extract valuable information from unstructured clinical notes to.
• Ingested clinical notes into Azure Data Lake Storage using Synapse Pipelines.

Livonia, Michigan, United States

• Predicting Claim Expense (Health Insurance):Problem Statement: To develop an advanced predictive model in Amazon SageMaker to accurately forecast healthinsurance claim expenses over the duration of one-year.
• Model Development: Built image-based deep neural networks to predict expected claims cost over one-yearpost-renewal date. Developed two separate models for member-level and group-level predictions.
• Data Utilization: Utilized data at the claim line level, including all medical, procedure, diagnosis, drug codes, anddemographic information.
• Feature Engineering: Created embeddings for all medical, drug, and demographic features using TensorFlowlayers like Text Vectorization, String Lookup, and Embedding layers. Added custom Lambda layers for.
• Decision Support System: The solution acts as a decision support system for underwriters, aiding in better policypricing during renewals.Project: ESG Performance Analysis and Improvement ToolProblem Statement: Need for.
• Aggregate ESG-related data from diverse sources like MarketPsych, Refinitiv, Bloomberg, and company websites.

Houston, Texas, United States

• Project: Optimizing Delivery RoutesProblem Statement: Minimize fuel consumption and delivery time by optimizing delivery routes.Responsibilities:
• Developed a route optimization algorithm using techniques like genetic algorithms or simulated annealing.
• Integrated real-time traffic data and package delivery constraints into the optimization model.
• Evaluated the impact of route optimization on fuel consumption, delivery time, and customer satisfaction.Technologies:
• Python,NetworkX,Google Maps API,AWS LambdaOutcomes:
• Reduced fuel consumption by 10% and improved delivery time by 15% saving $1.2 million annually.

Chennai, Tamil Nadu, India

• Project: Customer SegmentationProblem Statement: Segment customers into distinct groups based on their behaviors and preferences to tailormarketing and product offerings.Responsibilities:
• Conducted customer segmentation using K-means clustering algorithms to identify distinct customer segments.
• Analyzed customer behavior, demographics, and transaction data to identify key segmentation variables.
• Developed targeted marketing campaigns for each customer segment.Technologies:
• Python,Scikit-learn,Tableau,AWS SageMaker, S3, EC2Outcomes:
• Increased marketing campaign effectiveness by delivering personalized messages to targeted customersegments.

Bengaluru, Karnataka, India

• Project: Fraud Detection in Healthcare ClaimsProblem Statement: Identify fraudulent healthcare claims to prevent financial losses and ensure the integrity of thehealthcare system.Responsibilities:
• Developed a fraud detection model using anomaly detection techniques (one-class SVM) to identify suspiciousclaims.
• Engineered features from claims data, provider information, and patient demographics.
• Evaluated model performance using metrics such as precision, recall, and F1-score.Technologies:Python,Scikit-learn,PyOD,AWS SageMakerOutcomes:
• Identified and prevented fraudulent claims, $1 millions of dollars for the company annually.
• Improved the efficiency and accuracy of claims processing.Project: Developing a Patient Risk Stratification ModelProblem Statement: Identify patients at high risk for adverse health events to enable early.