ngqd42

Directing a RM 1.94M National AI Portfolio (MOSTI/MOHE) under the 12th Malaysia Plan, transitioning a Westport pilot to a multi-site production rollout across Peninsular Malaysia. Architected a 5-pillar automation pipeline (Registry, Vision Inspection, Repair Estimation, Allocation, Routing) using VLM and Computer Vision, benchmarking a 73% throughput increase. Engineered a 1.1 labor-hour reduction per container cycle by scaling R&D to national operations via Kubernetes, Argo CI/CD, and TensorRT-optimized edge inference. Governing technical architecture and cross-functional teams for mission-critical logistics, ensuring 99.9% reliability through Core Isolation, CUDA optimization, and FastAPI/HMAC security.

Eliminated 100% of manual verification cycles for Binance ($76B/day volume) by engineering an enterprise-scale test automation framework that replaced the output of a 5-person manual QA team. Secured mission-critical trading workflows for 90M users by architecting a high-concurrency reliability pipeline integrated into CI/CD (GitHub Actions/Argo), reducing release latency to zero manual touchpoints.

Led Java programming lab sessions that contributed to 50% of the total assessment of Computer Science undergraduates.

Directed international R&D teams to top-tier finishes in global AI competitions, resulting in peer-reviewed publications in IEEE and a Q1 Elsevier journal (Impact Factor focus). Reduced manual data labeling overhead by 90% by developing a novel single-cell image segmentation method using Transfer Learning and Fine-Tuning, recognized at the SATU SMART Innovation Competition (NCKU).