Directed advanced applied AI research and end-to-end development of production-grade intelligent systems, deploying Computer Vision (CV), Large Language Models (LLMs), and Vision-Language Models (VLMs) within real-world fintech and payment infrastructures.

• Multimodal Architecture: Architected next-generation AI-powered payment ecosystems by integrating multimodal intelligence, enabling real-time decision-making across embedded banking devices and transaction platforms.
• Edge CV Pipelines: Designed high-performance vision-driven device intelligence systems for transaction monitoring, anomaly detection, and operational analytics — significantly improving detection accuracy, system throughput, and response latency.
• Process Automation: Led the integration of LLMs and VLMs into fintech workflows to automate complex processes, generate contextual insights, and produce intelligent reporting pipelines.
• Hardware Optimization: Engineered scalable, low-latency AI pipelines optimized for resource-constrained edge environments (POS terminals, ATMs), ensuring robust inference under strict VRAM/RAM limitations.
• Security AI: Advanced AI-driven security frameworks by incorporating behavioral modeling and multimodal anomaly detection.

Led the research, development, validation, and certification of mission-critical operating systems powering payment terminals and ATM networks across one of the largest banking infrastructures in the region.

• R&D Lifecycle Management: Directed full lifecycle R&D for payment terminals, ATMs, POS systems, cashless platforms, and biometric/identity devices.
• Security Architecture: Owned and evolved the Security Platform (SP) architecture, implementing secure boot chains, cryptographic key management, trusted execution, and system integrity enforcement.
• OS Certification: Architected Java-based OS-level validation frameworks covering functional, performance, security, and compliance testing. Led multi-platform certification across Prolin, Monitor, PayDroid, and Android.
• Global Vendor Integration: Defined hardware–software integration strategies for international vendors including PAX, NexGo, Kozen (XC Tech), Hyosung, GRG Banking, and LKS.
• Performance Tuning: Optimized system-level performance for embedded payment environments, achieving ultra-low latency and high reliability under heavy transaction workloads.

Selected for a highly competitive, graduate-level computational neuroscience program focused on the neural and algorithmic foundations of decision-making under uncertainty.

• Brain-Inspired AI: Investigated brain-wide representations of prior knowledge and uncertainty through probabilistic decision-making paradigms (leveraging International Brain Laboratory data).
• Probabilistic Modeling: Designed Bayesian computational models to infer latent cognitive variables underlying decision processes, linking observable behavior to hidden neural mechanisms.
• High-Dimensional Data: Applied machine learning to extract interpretable structure from Neuropixels electrophysiology and calcium imaging recordings.
• Pipeline Engineering: Engineered robust data processing pipelines for multimodal neural datasets (denoising, temporal alignment, cross-session validation).

Contributed to the development and validation of AI-driven control systems for autonomous heavy-duty vehicles and intelligent transportation platforms.

• Real-Time Execution: Integrated embedded AI components into vehicle control architectures, connecting perception outputs with decision-making layers under strict latency and functional safety constraints.
• ADAS Validation: Designed and validated autonomous driving controllers, ensuring compliance with commercial vehicle safety standards.
• Hardware Integration: Contributed to autonomy stacks through radar sensing integration, control modules, and in-vehicle edge computing.
• International Deployment: Supported product adaptation across Asia-Pacific markets in collaboration with JAC Motors, CNHTC, and AXera Technologies.

Designed AI-driven safety and driver-assistance systems for heavy-duty trucks and commercial vehicles.

• Implemented AI-enhanced control logic within resource-constrained embedded environments.
• Validated ADAS controllers aligned with automotive safety standards.
• Optimized low-latency, real-time execution on specialized automotive hardware.
• Performed end-to-end safety validation, robustness testing, and hardware-in-the-loop (HIL) benchmarking.

Core contributor to a championship-level humanoid soccer robotics team, contributing to RoboCup competition successes across Australia, Canada, Thailand, Russia, and Japan.

• Dynamic Perception: Designed real-time CV pipelines for object detection, ball tracking, and robot localization under noisy, adversarial conditions.
• Multi-Agent AI: Developed architectures supporting coordinated team strategies and distributed decision-making.
• Intelligent Control: Applied reinforcement learning for tactical behaviors and real-time posture recovery.
• Edge Optimization: Deployed lightweight AI models for ultra-low-latency inference directly on humanoid microcontrollers.