Bridging AI models with production-grade backends.

Singapore-based Bank-Tech SaaS company.

AI-driven enterprise platform uniting ticketing, predictive maintenance, knowledge management and an AI assistant on a locally hosted LLM.

accuracy detecting equipment failures 7–14 days ahead

~92%

per knowledge query, down from 30 minutes

<30 s

for purchase approvals, down from 21

3 days

• ML-powered predictive maintenance engine detecting equipment failures 7–14 days in advance with ~92% accuracy, reducing maintenance costs by 25–30% and cutting emergency fault resolution from 4–6 hours to ~15 minutes
• RAG-based AI assistant over the company's internal knowledge base with role-gated access control, cutting knowledge query time from 30 minutes to under 30 seconds
• Intelligent ticketing system with AI-driven auto-categorization, smart assignee recommendation (skill matrix, workload, historical success rate), and automated workflow routing — improving operational efficiency by 30%
• Automated equipment purchase approval workflow, reducing cycle time from 21 days to 3 days with AI-assisted form auto-population and multi-level real-time tracking — eliminating data-entry errors entirely
• Multi-role user architecture (CEO, Manager, Technician, Engineer, etc.) with differentiated AI responses and permission-scoped data access across all platform modules

Fine-tuned Malaysia travel agent on Vertex AI, built with LangGraph and n8n. APICTA 2025 national finalist.