Contributed to complex AI training and evaluation pipelines, adapting quickly to new codebases across diverse 2-3 month project rotations:

• Agent Training & Production Debugging: Resolved complex production issues within major open-source repositories. Documented end-to-end problem-solving processes to generate high-quality behavioral data. Enforced strict quality control by benchmarking all solutions against historically merged PRs, ensuring output matched or exceeded original production-grade standards.

• Adversarial Challenge Design: Engineered adversarial coding challenges to intentionally expose AI model limitations and architectural edge cases. Developed comprehensive solutions and documented system failure points across multiple dedicated tech stacks, including Python, FastAPI, JavaScript, React, Rust, and SQL

• Developed IoT parking management system deployed across 12 mall locations using Python and Raspberry Pi

• Optimized gate response time from 4 seconds to instant through multithreaded programming

• Automated deployment pipeline with bash scripting to eliminate manual processes and version drift

• Configured CentOS servers and integrated PDF reporting into production PHP application