cv

Software Development Engineer Intern - RDS Proxy Team

Amazon Web Services (AWS)

Developed performance testing and visualization tools for RDS Proxy deployments

• Developed IPEBench Data Visualization Platform with Streamlit, creating 8 interactive visualization types that reduced engineers' performance regression analysis time from 8 hours to 15 minutes
• Built production-grade Regression Testing Framework (10,000+ lines) with statistical analysis (Welch's t-test, power analysis, Bonferroni correction) achieving 99% confidence in detecting performance regressions
• Implemented adaptive sampling system using Thompson Sampling and Bayesian optimization for convergence detection, improving test reliability from 47% to 90% and eliminating false positives that previously blocked deployments
• Integrated AWS CloudWatch metrics with automated dashboard generation, enabling real-time performance monitoring and data-driven decision making for RDS proxy deployments across multiple regions

Graduate Research Assistant

RobotiXX Lab, George Mason University

Developing Grounded Curriculum Learning for efficient Reinforcement Learning in Robotics

• Developed Grounded Curriculum Learning (GCL), integrating real-world data with adaptive simulated task generation
• Achieved 24.58% higher success rate and 50% improved sample efficiency
• Paper submitted to IEEE International Conference on Robotics and Automation (ICRA) 2024

Research Assistant

Computational Engineering and Robotics Lab, CMU

3D AR Scene Inpainting via Deep Learning

• Developed end-to-end deep learning pipeline achieving 92% accuracy in scene completion
• Implemented GAN model improving texture realism by 35% over baseline
• Applied RANSAC and DBSCAN algorithms reducing processing time by 40%

Ph.D.

George Mason University

Computer Science - AI and Robotics

• Advanced Machine Learning
• Deep Learning
• Reinforcement Learning
• Computer Vision

M.Sc.

Carnegie Mellon University

Mechanical Engineering

• Machine Learning
• Deep Learning
• Computer Vision
• Deep Reinforcement Learning & Control

Flexible Long-Term Mortality Prediction

Developed survival analysis model integrating CNN with Cox Proportional Hazards model

• Integrated MobileNet v2 with Cox Proportional Hazards model
• Implemented attention mechanisms for critical area focus
• Achieved 15% improvement over traditional methods