Objective, tools, measured impact — for every project. Transformation programmes and governance work come from regulated banking environments; AI and agentic projects span professional and Master's work.
Steer a time-critical, multi-vendor SaaS core-banking migration in private banking without interrupting operations — parallel-run and transition scenarios designed to make failure survivable before go-live.
Tools & methodsRequirements engineering · technical business analysis · PoC design · implementation blueprints · parallel-run architecture
0 downtime at go-live · 20%+ external cost reductionGive a private bank a defensible way to say yes to AI: co-developed the enterprise framework on the EU AI Act, NIST AI RMF and ISO/IEC 42001, then stress-tested it against real use cases — including RAG-based compliance assistants — for risk, transparency and documentation.
Tools & methodsEU AI Act · NIST AI RMF · ISO/IEC 42001 · use-case risk assessment · control design
Board-level recommendations, adopted · audit-ready controlsTurn the Change Advisory Board from a meeting into a system: digitised and operationalised CAB workflows, established decision structures for change, incident and service governance, and delivered data-driven SLA oversight for C-level reporting.
Tools & methodsITIL · change / incident / service governance · SLA analytics · GRC audit cycles
Audit resilience by design · standing C-level reportingPrepare a bank to classify, escalate and formally report major ICT incidents in line with DORA and its technical standards — governance and organisational readiness, not just a policy document.
Tools & methodsDORA · incident classification & escalation design · reporting readiness · regulatory analysis
Formal reporting readiness · embedded in operationsMake outsourcing reversible: developed cloud exit strategies, contributed to operating and control model design, and led third-party risk assessments across transformation and outsourcing initiatives.
Tools & methodsExit-strategy design · third-party risk assessment · operating & control models · outsourcing governance
Exit paths documented before dependency, not afterCut the manual overhead out of regulated workflows: orchestrated cross-platform low-code automation across M365 and Jira, with HCI-informed enablement so adoption survived contact with real users.
Tools & methodsPower Automate · Jira automation · low-code workflow design · HCI enablement · change adoption
60% fewer manual steps · sustained user uptakeMaster's data-science track: building and evaluating ML models with ensemble methods, with the emphasis on honest evaluation — validation design, metrics selection, and knowing when a model should not ship.
Tools & methodsPython · scikit-learn · ensemble methods · model evaluation · XAI / explainability
Evaluation-first practice, carried into AI governance workThe transitional moment: a governed process application composed entirely without custom code, built during the Master's programme. It reframed the question from “how do we code this?” to “what happens when the components can reason?” — the origin of the agentic systems work on this site.
Tools & methodsNo-code / low-code platform design · process modelling · governed workflow composition
The pivot into agentic AI — see the Agents pageA live demonstration rather than a claim: static frontend on GitHub Pages, serverless edge on Cloudflare Workers, and a CrewAI multi-agent pipeline feeding the dashboard — all dependency-light, WCAG AA, and governed by the same discipline as the day job.
Tools & methodsVanilla JS · Canvas 2D · CrewAI (Python) · Cloudflare Workers · GitHub Actions
You are looking at it — dashboard & agents pages show it liveBanking programme work is described at the level confidentiality allows — detailed case walkthroughs available in conversation.