ANNEX B · FIELD-NOTES-PB-001Available · Brisbane / Remote
PORTFOLIOSelected Work3 engagements

Solution Architecture · AI Platforms

Portfolio .

Three programmes where I owned the work end-to-end as Developer, Designer and Architect. Each pairs a high-level overview with a solution design, drawn with the standard Solution Design Viewer used across this site.

P-01Recusant Intelligence — MVPDeveloper · Designer · Architect
P-01 · HEROMonochrome
Fig. P-01 — Recusant Intelligence MVP, AI Solution Architect.
Fig. P-01 — Recusant Intelligence MVP, AI Solution Architect.
DeveloperDesignerArchitect
Next.jsReact.NETAzure OpenAILangChainMCPTypeScriptAzure AI Search

Recusant Intelligence is an AI-native platform delivered as a minimum viable product. As AI Solution Architect I owned the architecture from discovery through to a shippable MVP: a Next.js front end over a .NET API layer, with an MCP-based integration boundary that lets agents call external systems safely.

The design centres on a LangChain orchestrator that retrieves from an Azure AI Search vector store and generates with Azure OpenAI, wrapped in structured observability so every interaction is traceable.

P-01.1Solution Architecture

Rendering diagram…

Fig. P-01.1 — High-level Recusant Intelligence solution architecture (MVP).
GitHub Repository ↗Live Site ↗Field Notes →
P-02AgentsOS — Personal ProjectDeveloper · Designer · Architect
P-02 · HEROMonochrome
Fig. P-02 — AgentsOS, a personal agent-orchestration platform.
Fig. P-02 — AgentsOS, a personal agent-orchestration platform.
DeveloperDesignerArchitect
Next.jsReactTypeScriptNode.jsOpenAIVercelTailwindAgent Orchestration

AgentsOS is a personal research platform for orchestrating autonomous agents. I designed and built the full stack: a Next.js chat and agent-builder experience over a Node.js agent service, with a planner, memory and a tool registry that workers call at runtime.

Below are low-fidelity wireframes from the design phase — the chat surface, the visual agent builder, the conversation view and the integrations settings — used to validate UX intent before build.

P-02.1Solution Architecture

Rendering diagram…

Fig. P-02.1 — AgentsOS solution architecture: planner, memory, tool registry and streaming.
Fig. P-02.2 — Low-fidelity wireframes (design phase)
  • Integrations — tool and connection settings

    Integrations — tool and connection settings

  • Agent Builder — visual orchestration canvas

    Agent Builder — visual orchestration canvas

  • Chat — conversational agent interface

    Chat — conversational agent interface

  • Integrations — tool and connection settings

    Integrations — tool and connection settings

GitHub Repository ↗Field Notes →
P-03CeCiL — Clause IntelligenceDeveloper · Designer · Architect
P-03 · HEROMonochrome
Fig. P-03 — CeCiL, an AI-powered clause intelligence platform.
Fig. P-03 — CeCiL, an AI-powered clause intelligence platform.
DeveloperDesignerArchitect
Azure OpenAIRAGLangChain.NETReactTypeScriptAzure AI SearchCompliance

CeCiL demonstrates applied AI for the legal domain: a retrieval-augmented platform that turns a clause library into a queryable intelligence layer. As Developer, Designer and Architect I delivered the RAG pipeline, the compliance guardrails and the React experience.

Documents are parsed and chunked, embedded with Azure OpenAI, and stored in an Azure AI Search vector index. Queries are embedded, reranked and grounded against the index before Azure OpenAI returns a cited answer under PII and compliance checks.

P-03.1Solution Architecture

Rendering diagram…

Fig. P-03.1 — CeCiL RAG architecture with compliance guardrails.
Back to Portfolio →Field Notes →