Expertise

Agent systems built for real operations

I build agent systems as production infrastructure that can be deployed, observed, governed, and improved over time. The work spans multi-agent orchestration, model gateways, runtime packaging, control surfaces, approval paths, observability, and domain-specific workflows.

About EmmanuelView related projects

What the work covers

Agent systems need orchestration, runtime control, approvals, tool access, and observability around behavior in production. The work covers the stack needed to make automation safe, visible, portable, and useful in real operations.

  • Agent workflows deployed on Kubernetes and other production infrastructure environments
  • Multi-agent orchestration, specialist routing, and domain-specific research agents
  • OpenAI-compatible and Ollama-facing gateway layers, plus runtime control surfaces
  • Observability, approval loops, health checks, and outcome tracking

Where I’ve built this

Rusha-Corp and TierraViva work includes self-registering multi-agent systems, GitOps-backed agent configuration, proxy and gateway infrastructure, biodiversity navigation profiles, deep research agents, and analytical assistants grounded in live data products.

Tools I use here

Each tool below has a specific role inside agent systems. Fullstack delivery tools (marked) are how this domain’s own UIs, dashboards, and APIs get built.

Tools

Hermes — agentic runtime frameworkNATS JetStream — agent state and specialist routingLLM (OpenAI, Ollama) — model gateway and reasoning coreSlack API — human-in-the-loop approval workflowsKubernetes — agent deployment and schedulingKEDA — event-driven agent autoscalingPython — agent and runtime implementationFastAPI · React · TypeScript — fullstack delivery (agent control surfaces, profiles, dashboards)

Related projects

agent systemsai agentsagentic frameworkskubernetesoperational aillm workflows

AI

Agentic Systems

Production agent systems spanning multi-agent orchestration, domain-specific research agents, runtime-backed workflows, profile distributions, and model-gateway infrastructure. The work includes Kubernetes-based deployments, NATS-backed specialist routing, OpenAI-compatible and Ollama-facing gateway layers, approval paths, observability, and domain workflows for biodiversity, policy, and analytical operations, with Hermes used as one runtime within that stack. Serves the Agent Systems domain.