Graph8
Advisor / CTO role at Graph8. Led AI development and data engineering for an enterprise platform combining campaign orchestration, infra mapping, and agentic automation through a unified MCP architecture.
Key metrics
Architecture
Three-tier graph-native architecture. Frontend layer of React component libraries, visualisation engines, and an AI assistant panel. Backend layer of FastAPI + LangChain + MCP-based agent orchestration with task queuing. Domain-specific services for campaign management, software architecture planning, and cloud infrastructure ingestion.
Case study
Graph8
Advisor / Chief Technology Officer — Dec 2024 – Jun 2025.
Graph8 is an AI-driven enterprise platform for automating complex workflows and visualising system architecture. It combines campaign management, infrastructure mapping, and agentic automation into a unified, graph-native system, built around the Model Context Protocol.
Role
Joined as Advisor and stepped into the CTO seat. Led AI development and data engineering, set the technical direction for the platform's MCP-based agent architecture, and guided the team through the buildout of the core component ecosystem.
What Graph8 actually is
Graph8 is built as a coordinated set of components rather than a monolith — graph-native by design, MCP-first for tool orchestration. The pieces that make up the platform fall into three layers:
Frontend layer (TypeScript / React)
- g8-components — configurable React component library backed by REST/GraphQL/PostgreSQL
- graph8-components — Bit.dev + Storybook component publishing
- graph-viz — D3-based system architecture visualisation with hierarchical layouts
- g8wf — event-driven workflow engine (Jitsu events → REST actions)
- graph8-assistant — LangChain agent UI panel with Activepieces integrations
- graphql8 — sales analytics dashboard (GraphQL + Hasura + AG Charts)
Backend / agent layer (Python)
- g8fastapi — FastAPI + MCP tools + LangChain agents + Taskiq/Redis background jobs
Domain-specific services (Python)
- g8_mcp — MCP server for campaign management, A/B testing, audience targeting, metrics
- graph-coder — graph-based software architecture planning via LangGraph workflows
- cartographer — multi-cloud infrastructure ingestion to Neo4j
Tech stack
Python, FastAPI, LangChain, MCP, Taskiq, Redis on the backend. TypeScript, React, D3.js, Storybook on the frontend. Neo4j, PostgreSQL, GraphQL/Hasura for data. Deployed via Docker Compose.