I build and lead AI, data, and financial platforms in regulated environments. Across more than 15 years, I have worked in product management, software engineering, platform architecture, data systems, and technology leadership.
I currently lead technology for a cloud-native Islamic Banking-as-a-Service platform at Maybank Islamic. Previously, as VP of Product and Data Platform at LinkAja, I built the data and AI capabilities behind a payments platform serving more than 12 million verified users, including e-KYC, fraud detection, credit scoring, and LLM-powered access to company data. I hold a BEng in Electrical Engineering from Universitas Brawijaya and have studied Artificial Intelligence at the graduate level at Universität zu Lübeck, Germany.
- AI systems and orchestration: multi-agent systems for executive and enterprise use, with always-on workflows, controlled tool access, and multi-model routing with automatic fallback.
- Data platforms and generative AI: cloud data architecture, pipeline orchestration, governance, and secure LLM access to company data.
- Retrieval-augmented generation: retrieval strategy, reranking, and evaluation of RAG systems.
- Advisory: AI and data platform strategy, architecture, and delivery for regulated industries. I take on a small number of consulting engagements.
Languages: 
AI and agents: 
Retrieval and RAG: 
Data and cloud: 
BI and analytics: 
Dev and delivery: 
I contribute to open-source projects in areas I use in my own work, including agentic AI frameworks, computer vision and dataset tooling, and developer and localization tooling.
I use Claude Code and Codex as part of my engineering practice for understanding codebases, implementing focused changes, testing assumptions, and reviewing the result.
AI assistance does not remove accountability. I remain responsible for problem framing, technical constraints, verification, and every change I submit.
I approach product work the same way: connect user and commercial outcomes to architecture, data, delivery risk, and regulatory requirements, while staying close enough to implementation to challenge assumptions.