Lead Platform & Systems Integration Engineer with deep expertise in cloud-native infrastructure, multi-agent AI orchestration, and mission-critical enterprise systems integration. Proven track record in orchestrating multi-node distributed Kubernetes (K3s) clusters over mesh networks, architecting in-cluster autonomous CI/CD pipelines, engineering intelligent document extraction systems (OCR & vision LLMs), and delivering enterprise-grade middleware connecting Tier-1 ERPs (Oracle NetSuite, Odoo), collaborative workspaces (Lark Base), and financial engines across 50+ production microservices.
- Problem & Architecture: Manual entry of multi-page, non-standardized supplier quotes and complex Purchase Orders (POs) was prone to human error and created operational bottlenecks.
- Solution: Engineered an automated document intelligence pipeline combining high-precision computer vision, OCR preprocessing, and multimodal LLMs.
- Outcome: Converts unstructured PDFs and image scans into normalized, validated tabular data with line-item tax and SKU reconciliation. Seamlessly streams data directly into Lark Base tables and enterprise procurement ERPs, reducing manual data entry time by over 85%.
- Production Line Control: Architected a custom shopfloor MES for a leading commercial seafood processing enterprise. Integrated barcode weighing stations, cold-chain batch tracking, and real-time yield monitoring directly with an on-premise Odoo ERP backend.
- Industrial Formulation ERP: Designed manufacturing execution modules for high-precision aluminium tube fabrication, managing multi-stage extrusion scheduling, raw alloy formulation, quality inspection gates, and inventory lot traceability.
- High-Rise Commercial Facility Platform: Designed and built an end-to-end 8-module BMS platform utilizing Next.js 16 + PostgreSQL 18.
- Capabilities: Encompasses automated tenant tenancy lifecycles, tiered utility billing calculation, preventive maintenance work orders, IoT sensor anomaly alarms, visitor gate clearance, and facility asset depreciation tracking.
- High-Value Equipment Asset Tracking: Developed an integrated WMS and Asset Management System for a tier-1 distributor of premium commercial fitness equipment (Technogym partner).
- Features: Serialized asset lifecycle tracking, pre-delivery inspection (PDI) workflows, multi-warehouse transfer reconciliation, and warranty service dispatching.
- Intelligent Expense Parsing: Engineered an automated receipt and expense extraction engine (
kwix) capable of handling crumpled thermal paper, receipts, and invoices. - Features: Extracts line items, merchant tax numbers, and dates while performing automated anomaly detection, currency conversion, and duplicate expense identification.
- Mission-Critical Financial Bridge: Built an enterprise integration gateway (
netspark) bridging transactional banking feeds, billing events, and e-commerce systems with Oracle NetSuite ERP. - Reliability: Guarantees zero financial data loss through transactional idempotency, automated dead-letter retry queues, and end-to-end reconciliation logging.
- Legal Compliance Workflow: Developed middleware (
virtuesign) integrating workspace databases (Lark Base) with official digital signature providers (Mekari Sign & PERURI e-meterai). - Impact: Fully automated contract generation, legal stamp dispatch, signatory routing, and secure storage of government-certified signed PDF documents.
- On-Device Kiosk Attendance: Deployed an edge computer vision kiosk for retail branch employees featuring anti-spoofing liveness detection and rapid facial recognition.
- Integration: Feeds verified clock-in/out timestamps in real time into payroll and attendance scheduling engines with zero manual timesheet audits.
- Cloud-Native Topology: Orchestrated a 3-node distributed Kubernetes (K3s v1.36) cluster spanning cloud VPS nodes inter-connected via Tailscale Mesh VPN (
100.x.x.x). - Runner Fleet & Registry: Deployed org-scoped in-cluster GitHub Actions self-hosted runners (
build-systemnamespace) paired with an internal container registry. Completely bypassed third-party hosted runner minute limits and sped up Docker build rollouts by 4x. - Observability: Implemented lightweight log aggregation using VictoriaLogs + Vector/Fluentbit (<100MB RAM footprint), replacing bulky ELK stacks.
- AI Agent Coordination: Created
pm-orchestrator, a persistent VPS background orchestrator coordinating multiple LLM agents (Claude Code, OpenAI Codex, Google Antigravity) inside dedicatedtmuxsessions across 50+ repositories. - Custom MCP Servers: Standardized Model Context Protocol (MCP) servers enabling LLM agents to safely interact with Lark Suite OpenAPIs, Cloudflare edge DNS, and backend frameworks.
- DevEx Persistence: Built remote headless IDE terminal hooks with automatic session resurrection, preserving live conversational context across network reboots.
- Regional Marketplace Platform: Architected a multi-tier e-commerce engine connecting regional MSMEs (UMKM), hotel hospitality chains, and food producers in Eastern Indonesia (Flores Timur).
- Features: Multi-vendor catalogue, automated inventory sync, payment gateway integration, and localized logistics dispatch.
- Domain-Specific AI Chatbots: Engineered conversational legal and tax compliance assistants grounded with Indonesian legal corpus and tax calculation logic for corporate advisory.
- Cloud & Orchestration: Kubernetes (K3s), Docker, Tailscale Mesh VPN, Cloudflare (DNS, Workers, Edge WAF), Nginx, VictoriaLogs, Vector, Prometheus.
- Backend & Middleware: TypeScript, Node.js, Go, Python, PHP, Laravel, FastAPI, REST APIs, Webhooks, SSE, gRPC.
- Frontend & UI/UX: Next.js (App Router), React, Tailwind CSS, shadcn/ui, Radix UI, Framer Motion, Responsive & Accessible Design (WCAG 2.1 AA).
- Databases & Storage: PostgreSQL, MySQL, Redis, MinIO S3 Object Storage, SQLite.
- Enterprise Platforms & APIs: Oracle NetSuite ERP, Odoo ERP, Lark Suite (OpenAPI, Base, Bots), WhatsApp Business API, Mekari Sign.
- AI & Machine Learning: Model Context Protocol (MCP), Multi-Agent Systems, Ollama, LiteLLM, Document OCR/Vision LLMs, LangChain/LlamaIndex paradigms.
