20+ years untangling enterprise complexity across Ecom, Retail, Healthcare, Banking, Telecom and Pharma. These days I’m focused on turning that experience into AI systems and agents that can do useful work in real-world enterprise contexts.
Explainable GenAI career intelligence for vacancy matching and career direction discovery
GitHub repository Live product architecture site
Sweepster is a product architecture case study for an AI-native vacancy matching system. It turns messy job descriptions into explainable, scored matches against a candidate capability profile, showing why a role matches, why not higher, what is a real gap, and what is still unknown.It demonstrates the kind of AI systems work I’m moving towards: agentic workflow design, LLM-assisted requirement extraction, modular pipeline architecture, no-imputation logic, human-in-the-loop questions, reversible feedback learning, and schema-first implementation thinking.
AI-powered interview preparation agent
A Telegram bot + web dashboard that simulates real job interviews using your actual CV and job description. Personalised questions, structured feedback, Executive Language scoring, session analytics. Built on Python + Claude API. Not a prototype — a working product I use myself.
I've shipped products at scale with teams of 110+ across distributed programs — BT Group, Primark, AIB, Janssen, Bosch, Royal Mail, Henry Schein.
What's changed: I now integrate AI into how I work, not just what I build. On my last engagement (Primark, MarTech discovery) I used GenAI to run extended capability research and compress a 6-week analysis into days.
What hasn't changed: I still think the most important skill in product is knowing which problem is actually worth solving.
Python Claude API SQLite Telegram SAP Commerce Commercetools Salesforce Shopify Sitecore AEM Google Analytics Figma Miro Jira Confluence
