• Architected enterprise-grade conversational AI safety systems using NeMo Guardrails, implementing robust content moderation frameworks that reduced potential risks by 85% in production LLM applications • Led the adoption of NVIDIA AI Foundry platform for custom LLM development using Llama 3.1 architecture, achieving 40% improvement in model performance and deployment efficiency • Developed automated testing frameworks for LLM applications, ensuring consistent compliance with ethical guidelines while maintaining 99.9% system reliability

• Led end-to-end automation framework development for Ford's infotainment systems (Phoenix, Sync3), resulting in 70% reduction in testing cycles and 85% increased test coverage • Engineered comprehensive test suites for critical vehicle modules (Parking, RVC, Towing, Climate) on Android platforms, achieving 99% detection rate for system vulnerabilities • Developed and optimized Python-based API testing frameworks and debugging tools, reducing script execution time by 40% while expanding test coverage across multiple vehicle segments

• Architected automated picture quality validation system for Dolby Vision PCs using X-Rite device integration, reducing testing time by 65% and improving accuracy to 99.8% • Led software reference design development for Qualcomm chipset platforms (HQX), delivering technical solutions to key OEM partners and reducing integration issues by 70% • Developed and optimized cross-platform test automation framework, achieving 90% test coverage across multiple Software Products while maintaining 99% reliability rate

• Architected and deployed production-ready AI solutions for LG Mobile's Ellivision platform, integrating multi-modal recognition systems (scene, emotion, blur detection) with 95% accuracy in real-world conditions • Engineered optimized ML pipeline for facial landmark detection and image classification, reducing inference time by 40% while maintaining >90% accuracy across diverse datasets • Developed end-to-end data preprocessing and model training frameworks, resulting in 35% improvement in model performance and successful deployment across multiple LG mobile devices