JoshuaApollo

Built a custom AI agent that automates data synthesis from multiple unstructured sources, applying advanced prompt engineering and LLM orchestration to generate actionable analytical insights in real-time, analogous to building custom Gemini Gems for automated reporting.

Designed structured evaluation pipelines to benchmark output accuracy, iteratively refining prompt strategies to improve the reliability and precision of AI-generated analysis.

Delivered a polished end-to-end demo placing Top 8 at NTU's Fintech Innovators Hackathon 2026.

Built an AI-driven data processing pipeline that fuses multiple real-time data streams (vision, audio, physiological signals), automating classification and generating prioritised outputs within 10 seconds.

Designed the system as a replicable proof-of-concept with documented workflows, enabling rapid iteration and continuous improvement of data collection and analysis processes.

Proposed a novel energy-harvesting system that captures the piston-effect airflow generated by trains passing through underground MRT stations, using turbines to convert this otherwise-wasted tunnel wind into usable electrical energy.

Designed and built a working prototype that validated the concept's technical viability.

Designed a technology-driven product for healthcare, defining the product vision, technical architecture, and go-to-market strategy including SaaS pricing, market sizing, and a mapped procurement pathway.