Drefan Mardiawan

Organizations, governments, and communities often struggle to efficiently manage surplus resources, resulting in operational waste, fragmented distribution, and limited visibility across the resource supply chain.

Designed an enterprise-grade circular economy platform enabling businesses, governments, and communities to manage, redistribute, and optimize surplus resources through intelligent logistics, real-time matching, modular services, and scalable digital infrastructure built with Next.js, Node.js, and PostgreSQL.

Enabled organizations to improve resource utilization, reduce operational waste, strengthen sustainability initiatives, and support circular economy programs through a scalable enterprise platform.

Enterprise organizations require advanced RAG and LLM intelligence capabilities but face severe compliance, privacy, and latency risks when routing proprietary intelligence through third-party public AI APIs.

Designed a fully self-hosted Enterprise AI platform powered by a dedicated NVIDIA H200 GPU Cluster, integrating private Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), vector databases, and on-premise inference infrastructure for enterprise environments.

Reduced enterprise document analysis from hours to seconds through high-performance local inference while maintaining enterprise-grade security, privacy, and zero external AI dependencies.

Government services were fragmented across hundreds of independent systems, creating inconsistent citizen experiences, duplicated processes, and limited accessibility to public services.

Architected a province-wide Government Super App designed to unify thousands of public services into a single citizen-centric platform utilizing scalable backend services, centralized integration architecture, enterprise APIs, and infrastructure hosted within the Central Java Provincial Government Data Center.

Enabled citizens to access government services through a unified digital platform while supporting large-scale digital transformation, improving service accessibility, operational efficiency, and inter-agency collaboration across Central Java Province.

Commercial food and beverage enterprises struggle with fragmented POS terminal synchronization, order latency, inventory drift, and high transaction failure rates during peak dining hours.

Architected a cloud-native multi-tenant QR ordering and POS ecosystem with real-time inventory synchronization, strict ACID transactional guarantees, and automated offline-first fallback mechanisms.

Streamlined operational throughput across multi-location F&B chains, eliminated inventory drift, and reduced order processing overhead by 45%.