AI-native governance platform with a 9-dimension risk classification engine, 7 AI-native automated testing gates, and a compliance artifact generator producing GDPR DPIAs, AI Model Cards, and STRIDE Threat Models in 90 seconds — replacing 15 days of manual documentation.
Agentic AI in action — the full autonomous loop running live: Claude reads a Jira story, queries the Salesforce schema, creates the configuration via Metadata API, verifies via SOQL, posts a completion comment in Jira, and marks the story Done. Zero human intervention from start to finish. Built with Claude API + MCP protocol, this became the production pattern behind the open-source salesforce-ai-agent-mcp package.
Production-grade MCP server giving Claude direct access to Jira Cloud, Salesforce Tooling + Metadata APIs, and n8n workflows. Full agentic loop — reads Jira story, checks Salesforce schema, creates config, verifies via SOQL, posts Jira comment, marks Done. Zero human intervention.
MCP server connecting Claude to n8n healthcare automation — prior authorisation, patient task management, and clinical triage. Auditable human-in-the-loop checkpoints before any action. Claude's reasoning layer separated from n8n orchestration for isolated failure modes.
Privacy-first AI styling app powered by Claude vision. Upload a photo, select an occasion, receive 3 complete outfit recommendations with brands, prices, and shopping links. GDPR-compliant by design — no image ever stored. React + Express + Claude Vision + Supabase, deployed on Vercel + Railway.
Autonomous agent converting handwritten Product Owner notes into structured Jira Epics and Stories using GPT-4 Vision and LLM orchestration. Generates acceptance criteria, story point estimates, and dependency links directly in Jira — validating multimodal AI for delivery workflows.
MBA in Marketing & International Business · Bachelor of Commerce