Ardura
Broker CRM with AI-driven churn prediction, segmentation engine, and operator-controlled retention workflows.
Increased retention intervention quality while eliminating autonomous risk — every AI recommendation is operator-approved with full attribution.
Problem
Broker CRMs fail at retention because they either ignore AI signals entirely or pipe them straight to automation without operator judgment — creating liability when models drift and silent churn when they don't.
Role
- Sole engineer — CRM design through full-stack implementation
- Built segmentation engine, automation workflows, and operator approval queues
- Designed AI integration pattern: recommend → review → approve → trace
Approach
- Built a full CRM platform with client profiles, churn scoring, and segmentation
- Churn predictions surface as actionable recommendations, never trigger actions directly
- Segment → action loops: operators define rules, review predictions, approve interventions
- Human-in-the-loop: operators approve, modify, or reject every automation rule with full attribution
- Evaluation traces attached to every prediction — model confidence, input features, decision rationale
- Component reuse from Truvesta governance stack — safety invariants preserved across products
Outcomes
- Complete client lifecycle view: acquisition → engagement → risk → retention
- Zero autonomous actions — every retention intervention is operator-approved
- Auditable decision chain from churn signal to customer action
- Disciplined reuse across products without cutting governance
- End-to-end CRM: client profiles, segments, churn scores, retention actions
- Evaluation traces on every prediction: model confidence, input features, rationale
- AI predictions surfaced as recommendations — operators approve every action
- Reused governance components from Truvesta without cutting safety
What I refused to build
These constraints are part of the engineering signal: the work stayed useful because the unsafe shortcut paths stayed out.
- R-01Auto-execution of AI-driven retention actions — operator judgment is a hard requirement
- R-02Black-box predictions — every recommendation shows its reasoning
Stack
Hiring signal
This is the kind of work I want ASADI AI to make legible first: fintech systems where product direction, architecture, execution, and governance all have to line up.