From CMS platforms and admin dashboards to Discord AI gateways, OpenClaw runtimes, skills/plugins, and Kubernetes deployments — I build full-stack systems that actually ship.
Every system I ship moves through the same six layers. Scroll to watch an idea become live infrastructure.
Product / CMS
A CMS and SaaS platform where a team manages content, customers and AI-powered workflows — and where an assistant needs to actually do work, not just chat.
Backend / API
agent_api over GraphQL, MongoDB models and runtime diagnostics — the durable core that agent_ui and every runtime binds to.
Gateway
The Erxes AI Assistant Gateway authenticates, maps tenant → assistant → runtime, and routes every command to the right place.
Runtime / Skills
Managed OpenClaw runtimes provision on demand, search/install/enable skills and plugins at runtime, and keep sessions alive across long-running work.
Discord / Web UI
Discord slash commands and web dashboards — isolated threads, session continuity, and a clean admin wizard.
Cloud infrastructure
Hetzner, Docker and Kubernetes pods — consolidating fleets into orchestrated clusters that scale without runaway cost.
Across product, agents and infrastructure — the building blocks I ship most.
Content, customers, automation and AI workflows from one platform.
One shared bot, many tenants, the correct assistant every time.
Provision, health-check and operate assistant runtimes in the cloud.
Reasoning agents that use tools and install new capabilities.
OAuth, channel binding and thread-aware assistant sessions.
Containers, certs, scaling and cost — from code to production.
Cinematic, premium dark interfaces that actually ship.
Real SaaS infrastructure — gateways, runtimes and the cloud underneath. Not demos.
CMS, admin dashboards, SaaS workflows, plugin architecture, and enterprise product integration.
I work on CMS and SaaS systems that help teams manage content, customers, automation and AI-powered workflows from one platform — plugin-based, multi-tenant, and production-grade.
Plugin-based
Architecture
Multi-tenant
Tenancy
Admin UI
Surface
Multi-tenant Discord gateway connecting one shared bot to many SaaS AI assistants.
One official Discord bot routes /assistant commands to the correct assistant runtime per tenant — OAuth install, signature verification, tenant/guild/channel/thread mapping, session continuity, thread isolation and long-running jobs over secure runtime forwarding.
One shared
Bot
Per-tenant
Routing
Persistent
Sessions
Automated cloud provisioning system for OpenClaw assistant runtimes with health checks, diagnostics, and runtime management.
Provisions runtimes on Hetzner, approves runtime URLs, runs health checks, wires the OpenClaw gateway and adapters, propagates Kimi model auth, and exposes diagnostics plus skills/plugins management endpoints.
Automated
Provision
Verified
Health
Scalable
Hosting
Tool-using AI agents with runtime sessions, web/Discord parity, sub-agent workflows, and coding-agent style automation.
Agent systems that reason, use tools, install capabilities, connect to SaaS data and operate through web or Discord — runtime adapters, persistent sessions, sub-agents and chat-to-action workflows.
Web + Discord
Interface
Runtime
Tools
Chat-to-action
Workflows
Runtime capability system for searching, installing, enabling, inspecting, and managing assistant skills/plugins.
Capabilities aren't hardcoded — they're searched, installed, enabled, inspected and managed at runtime. Plugin doctor/inspect flows, ClawHub publishing and secure plugin handling let assistants gain new abilities safely.
Runtime
Lifecycle
ClawHub
Source
Secure
Handling
GraphQL APIs, agent_api, agent_ui, managed deployment wizard, diagnostics UI, and Discord connection flows.
agent_api and agent_ui wired through GraphQL queries and mutations — a managed assistant deployment wizard, runtime diagnostics UI, skills/plugins UI, Discord install/connect, channel binding and owner/admin flows.
GraphQL
APIs
Guided
Wizard
Owner/Admin
Flows
Moving from many separate assistant servers toward Kubernetes pods on large Hetzner machines for lower cost and better scale.
Consolidating fleets of OpenClaw servers into pod clusters on large Hetzner nodes — Docker, nginx, health checks and orchestration tuned for density, scale and predictable cost.
Lower
Cost
Higher
Density
Kept
Isolation
Cinematic landing pages, GSAP storytelling, 3D infrastructure maps, and premium dark SaaS interfaces.
Design-to-code cinematic web — 3D landing pages, GSAP scroll storytelling, animated infrastructure maps and premium UI prototypes. This site is one of them.
Cinematic
Style
Design-to-code
Build
GSAP + 3D
Motion
From a single user action to the right assistant runtime — every hop is mapped, secured and observable.
user → discord / web ui → gateway → tenant mapping → openclaw runtime → skills · plugins · tools
The work underneath — auth, routing, runtimes, and the deployment path to production.
OAuth flows
Install, consent and guild authorization.
Tenant / channel binding
Mapping tenants, guilds and channels to runtimes.
Runtime health verification
URL approval and live health checks.
Provider auth
Propagating model and gateway credentials securely.
Model configuration
Kimi setup, adapters and runtime config.
Secure forwarding
Signed, isolated request routing to runtimes.
GraphQL API design
agent_api queries and mutations for agent_ui.
Plugin / skill lifecycle
Search, install, enable, inspect, manage.
Docker / Kubernetes deployment
Containers and pods on Hetzner nodes.
Production debugging
Logs, certs, domains, runtime and auth failures.
I work end-to-end — the UI people touch and the orchestration that keeps it alive.
Cinematic, performant interfaces.
APIs and data that hold up under load.
Runtimes, tools and assistant orchestration.
From bare metal to orchestrated pods.
Systems that ship and sell.
Philosophy
Let’s build