Personalization that had to work everywhere at once
The platform needed real-time recommendation and personalization across all active markets — not a single-market experiment. That meant a system that stayed consistent while ML, backend, and data flows moved at different speeds, and a delivery plan that could get it to production without stalling on cross-team dependencies.
Run the AI build as a sequenced program
As program lead, I owned the sequencing and coordination rather than the modelling. I set the roadmap, aligned ML and backend engineering on shared milestones, and defined how OpenAI API-based services and n8n-automated data flows fit the production pipeline — then kept dependencies visible so no single team blocked the rollout.
- Translated a broad personalization goal into a sequenced, milestone-based roadmap.
- Coordinated ML and backend teams around one shared delivery plan.
- Scoped where OpenAI API-based services and n8n automation fit the real-time data path.
- Surfaced and resolved cross-team dependencies before they delayed a market.
From roadmap to a live system across markets
The program moved from roadmap through production as one coordinated effort — integrating OpenAI API-based services and n8n-automated data flows, and validating that the system held up in real time across every active market before it was considered shipped.
My value here was defensible AI delivery: I didn't train the models, I made a multi-team AI system ship on time and stay consistent across markets — tied to concrete tools, not buzzwords.
Live, real-time, everywhere
The result was a live, real-time personalization system running across all active markets, delivered as a coordinated program spanning ML, backend, and automated data flows — in production rather than in pilot.