Keerthana Senthil Raja

Data-driven professional with strong expertise in SQL, Python, and statistical modeling, specializing in transforming complex datasets into actionable insights. Experienced in data validation, predictive analytics, and machine learning, with hands-on work across healthcare, financial, and environmental datasets. Skilled in Power BI, data mining, and algorithm development to drive operational efficiency, automation, and informed decision-making. Known for strong problem-solving abilities, clear communication, and a collaborative, impact-focused approach.

Student Researcher, Seattle University, Seattle, WA 

• Applied data integration and transfer learning techniques to align and reconcile medical datasets from multiple healthcare domains, ensuring data quality and consistency across systems.
• Performed data validation, normalization, and error detection on 100+ patient records, simulating claims reconciliation and CMS submission workflows.
• Designed data mapping frameworks to improve interoperability between heterogeneous data sources, supporting accurate and compliant reporting.
• Utilized Python, SQL, and regression-based modeling to generate actionable insights for healthcare decision-making and operational improvements.
• Translated analytical findings into clear, stakeholder-ready summaries, promoting data transparency and informed clinical and administrative actions.

Risk Lens AI - Real time fraud detection System

• Designed and implemented a real-time financial fraud detection pipeline using streaming transaction data, enabling early risk identification and alert generation.
• Built and trained machine learning models for fraud classification with explainability features (e.g., feature attribution) to support transparent decision-making.
• Developed an end-to-end data pipeline integrating data ingestion, preprocessing, model inference, and structured storage for scalable analytics.
• Deployed the system using containerized microservices, ensuring modularity, reproducibility, and production-readiness.

AI Tutor - Adaptive Learning Assistant

• Built an adaptive AI tutoring system that personalizes explanations based on learner preferences (Visual, Auditory, Kinesthetic) to improve comprehension.
• Integrated educational datasets to classify student learning styles and dynamically adjust instructional strategies in real time.
• Implemented an LLM-powered response engine capable of generating text, visual, and audio-based explanations based on student feedback.
• Designed a modular backend architecture enabling continuous learning, assessment, and personalized content delivery.

AI Job Resumer - Resume Optimization & Job Matching Platform

• Developed an AI-driven resume optimization system that customizes resumes to match job descriptions while maintaining ATS compliance.
• Implemented NLP-based resume scoring and skill-matching algorithms to improve job-application relevance and ranking.
• Built an automated job recommendation engine that aligns candidate profiles with suitable roles across multiple job platforms.
• Addressed ethical considerations in resume modification by ensuring factual integrity, transparency, and bias-aware scoring mechanisms.

AWS Pipeline project

• Built a multi-module big data pipeline using Hadoop, Hive, Sqoop, and MapReduce to process airline, text, and log datasets.
• Developed custom MapReduce and HiveQL workflows for word count, TF-IDF computation, airline delay analysis, and log severity processing.
• Generated key insights, including Spirit Airlines' highest average arrival delay and higher vocabulary richness in Shakespeare's texts compared to Austen's.

Maze Solver with AI 

• Developed an intelligent maze-solving system using Bidirectional BFS and Q-Learning, combining deterministic and reinforcement-based navigation. Implemented dynamic path comparison, visualization, and logging to evaluate solution quality, exploration behavior, and algorithm performance.
• Conducted performance benchmarking to analyze optimality, convergence speed, and computational cost, highlighting key strengths and trade-offs.

Yoga posture detection using ML

• Developed a real-time yoga posture detection system using Google PoseNet for accurate skeletal tracking and pose estimation.
• Implemented image classification and pose-mapping algorithms to evaluate posture accuracy and identify deviations from ideal form.
• Built an end-to-end pipeline with instant feedback and model evaluation, enabling reliable posture assessment and safer practice.

Patient Health Monitoring System using IoT, ML

• Integrated SPO2, ECG, and temperature sensors with ML models to enable real-time patient monitoring, detecting physical and emotional health conditions for early diagnosis and timely intervention.
• Designed a scalable system with automated alerting, real-time data acquisition, and predictive analytics to ensure accurate, reliable, and rapid response to abnormal patient vitals.