AI Personalisation Product Lead

## AI Personalisation Product Lead Apply remote type:
Hybrid locations:
Johannesburg:
Cape Towntime type:
Full time posted on:
Posted Todayjob requisition :
JR-78933
** Let's Write Africa's Story Together!
** Old Mutual is a firm believer in the African opportunity and our diverse talent reflects this.
** Job Description
** The AI Personalisation Product Lead is the technical product owner for Old Mutual’s AI-driven personalisation engines. This is an engineering-adjacent leadership role that owns the product track for four interconnected AI systems: the Next Best Action (NBA) engine, the content personalisation engine, the dynamic optimisation infrastructure, and the chatbot personalisation capability. The Lead is accountable for the technical maturity, production reliability, and performance improvement of these systems — not for the commercial proposition they serve.
This is a deeply technical role. The ideal candidate has a background in machine learning engineering, data science leadership, or AI product management within a production ML environment. They must be fluent in the language of model training, feature engineering, serving infrastructure, latency optimisation, champion-challenger experimentation, and MLOps — not just able to discuss these topics at a conceptual level, but able to make architectural decisions, review model performance, challenge data science approaches, and unblock engineering bottlenecks.

This person runs sprint planning, reviews pull requests with the team, triages production incidents, and makes trade-off decisions between model accuracy and serving latency.  The role leads a technical team of 13 to 15 resources in data science, ML and engineering.##
** Key Responsibilities
** Next Best Action (NBA) Engine — Product Ownership
* Own the NBA product vision, technical roadmap, and engineering backlog — defining what capabilities the engine needs to develop, in what sequence, and to what performance standard.
* Define the NBA decision framework: what actions the engine evaluates, what signals it consumes, what constraints it respects (commercial rules, partner obligations, frequency caps, POPIA consent), and how it ranks competing actions.
* Direct the Data Science team on model development.
* Own the real-time serving infrastructure.
* Design and govern champion-challenger experimentation: define experiment protocols, statistical rigour standards, traffic allocation strategies, and decision criteria for model promotion.
* Build and maintain the feature store: curate the real-time and batch feature sets (behavioural, demographic, transactional, contextual) that feed NBA models, ensuring feature freshness, quality, and governance.
* Integrate NBA outputs into Rewards and Digital Platform customer touchpoints: work with Technology/Engineering on API design, integration patterns, fallback handling, and error management.
* Monitor NBA performance daily: model accuracy (AUC/precision/recall), acceptance rates, revenue attribution, latency metrics, and drift detection — triggering retraining when performance degrades.
* Build feedback loops: ensure action outcomes (accepted, ignored, rejected) are captured, pipeline-processed, and fed back into model training data on a defined cadence.
* Report NBA performance and attributed revenue to Business Owners monthly, with clear methodology transparency.

Content Personalisation Engine — Product Ownership
* Own the content personalisation product vision and technical strategy: how the engine determines which content, articles, offers, and recommendations to serve to each user.
* Direct the development of content recommendation algorithms: collaborative filtering, content-based filtering, hybrid approaches, and contextual re-ranking models.
* Build and maintain user interest profiles from behavioural data (content views, reads, dwell time, shares), stated preferences, and inferred affinities.
* Own the content taxonomy and tagging framework (jointly with Data Architect): define the metadata structure, classification schema, and machine-readable content attributes that enable effective algorithmic matching.
* Integrate personalised content feeds into…