I build and manage AWS cloud systems. I keep servers running, scale them automatically when traffic spikes, and automate routine tasks with Linux and Python.
Computer Engineering graduate from Universitas Andalas. Experienced in physical network operations at PT Semen Padang, automated AWS cloud setups, and Linux systems administration.
FEATURED PROJECT 01 // MULTI-TIER RESILIENT CLOUD ARCHITECTURE
PRODUCTION SPEC
AWS HIGH-AVAILABILITY SYSTEM
Resilient Multi-Tier Cloud Architecture
Dual-AZ Fault-Tolerant Web and Database Topology
GOAL & SCENARIO
Keep a web application online 24/7 across two AWS data centers without single-point-of-failure risks.
HOW IT WORKS
Traffic enters through an Application Load Balancer. Compute nodes scale on demand in private subnets. The database syncs in real-time to an automatic standby replica.
RESILIENCE TEST RESULTSVERIFIED LOGS
1.
Primary DB Crash Test:Simulated a database outage. Standby replica took over in 52 seconds with zero data loss.
2.
Traffic Spike Test:Triggered 75% CPU load. Auto-scaling launched a healthy new server in 84 seconds.
FINOPS & RESOURCE EFFICIENCY< $15/MO RUNTIME
Automated off-hours shutdown scripts and right-sized instances to keep operational costs under $15 per month.
FEATURED PROJECT 02 // AUTONOMOUS AI AGENT & ZERO-TRUST GATEWAY
ZERO-TRUST & AI INFRA
AWS LIGHTSAIL & AI GATEWAY
Autonomous AI Agent on AWS Lightsail
Zero-Trust Mesh Network, Multi-LLM Gateway and Telegram Daemon
GOAL & SCENARIO
Run an autonomous AI task runner 24/7 on an inexpensive VPS without opening web ports to the public internet.
HOW IT WORKS
The AWS firewall blocks public traffic. Admin dashboards are reached via an encrypted Tailscale WireGuard mesh tunnel. Tasks are sent through Telegram using outbound HTTPS polling.
Hairpin NAT Reflection Fix:Closing external ports caused public IP calls to time out. Fixed by pointing agent traffic directly to localhost loopback 127.0.0.1.
2.
Reliability & Resource Tuning:Configured 2 GB swap to prevent OOM kernel kills. Enabled systemd user lingering to keep the agent daemon alive across reboots.
FINOPS & RUNTIME COST$12/MO FLAT PREDICTABLE
Flat $12/month Lightsail instance with 3 TB transfer. Free-tier LLM pooling via OmniRoute with automatic rate-limit fallback keeps AI inference cost at $0.
TECHNOLOGIES & TOOLS:
AWS Lightsail Ubuntu Linux Docker Tailscale WireGuard Hermes Agent OmniRoute Telegram Bot API Systemd
TOPOLOGY // ZERO-TRUST AI ARCHITECTURE
Tailscale WireGuard Mesh Tunnel
Hermes Autonomous Agent Daemon
LIVE CAPTURE
Autonomous Telegram Execution
FULL RES ↗
Direct shell execution (`uptime`, `free -h`) via Telegram bot polling without human intervention.
STATUS:24/7 AUTONOMOUS SYSTEMD DAEMON ON AWS LIGHTSAIL