Onaolapo Odunjo

Forward-deployed production AI for regulated healthcare and fintech clients: map live operator workflows, make model and infrastructure tradeoffs, ship compound AI architectures with engineers, and drive adoption from launch through operator trust.

• ▸Operate as a broader partner to the CEO beyond the product mandate: pipeline, CRM, client conversations, weekly operating cadence, financial planning, investor reporting, unit economics, and company-level priorities.
• ▸Raised internal movement completion 34% → 72% at multi-site hospital systems: embedded with clinical and ops teams to map intake-to-discharge handoff failures, co-designed agent-assisted workflows with engineers, then owned adoption post-launch by running weekly override reviews, iterating prompts and routing, and tightening escalation paths until usage held.
• ▸Improved precision 22% → 50% and cut false positives 60% → 15% for a top 10 U.S. bank compliance team (~$600K annual savings): embedded 2+ days/week with fraud and risk analysts, translated live SOPs and policy boundaries into model requirements and eval cases, shipped compound agents with confidence gating and mandatory human review.
• ▸Set the AI quality bar governing what went to production, built evaluation infrastructure with 300+ labeled examples across fraud, compliance, and financial exploitation risk categories; established confusion-matrix benchmarks, precision/recall thresholds, and online eval from analyst overrides as hard promotion gates.
• ▸Architected nine-node agentic workflow (ingestion, extraction, enrichment, classification, anomaly detection, state management, confidence scoring, policy gating, human review) with Claude/OpenAI APIs and n8n; partnered with forward-deployed engineers on Python/FastAPI and Next.js delivery.
• ▸Designed HITL control layer exposing confidence scores, source evidence, escalation paths, and override capture to investigators, made AI outputs explainable, auditable, and subordinate to compliance judgment, not black-box automation.
• ▸Built an internal GTM Intelligence Platform (HAL) monitoring 57 market feeds: ranks relevant market signals, extracts named prospects, surfaces outreach opportunities, and gives leadership a live view of GTM activity and blockers.
• ▸Designed four-tier GTM intelligence system (Signal Capture → Enrichment → Intelligence → Execution); centralized ICP scoring in Claude-powered governance layer connected to Clay, HeyReach, and HubSpot; reduced GTM stack cost ~$215K annually.
• ▸Built enrichment architecture for a ~1,200-contact advisor segment: market scoring, geographic segmentation, company enrichment, social activity signals, and controlled CRM writeback.

• ▸Lead product direction and technical delivery for Kova, a financial operations and credit infrastructure platform spanning collections, payments, savings, and financing across web, WhatsApp, and partner integrations.
• ▸Built organisation onboarding, bank-statement imports, payment matching with evidence and review, instalment plans, reminders, and household account history.
• ▸Built an assistant that answers from permitted records and prepares actions for staff confirmation; connected Paystack payment adapters, partner APIs, and a replayable TypeScript business engine.
• ▸Designed a shared engine for school fees, rent collection, clinic bills, savings networks, and employer-funded advances; connected explainable Kova Scores with affordability checks, financing offers, and repayment workflows.

• ▸Owned product strategy for $148M ARR Prime subscription initiative serving 1.17M+ monthly active users within broader $8.4B business, translating churn analysis into activation, personalization, and lifecycle engagement roadmap.
• ▸Shipped ML-driven targeting framework for subscription personalization and retention, partnering with data science to define behavioral segments, measurement approach, and incremental lift analysis.
• ▸Aligned 8 cross-functional stakeholder teams (Data Science, Engineering, Marketing, Finance, and Legal) on targeting framework scope, success metrics, and launch criteria; presented roadmap to VP-level leadership.

• ▸Managed 13-person cross-functional platform team across PMs, engineers, BSAs, and data modelers; set roadmap priorities, operating cadence, requirements standards, and delivery accountability across 15 business units.
• ▸Built product strategy and investment case for $500M+ enterprise data platform by quantifying EBITDA impact of data reliability, compliance infrastructure, ML enablement, platform reuse, and capability sequencing; coordinated 170+ contributors.
• ▸Led 0→1 Azure data platform build after diagnosing structural fragmentation across source systems, metadata, and model inputs; defined canonical data model, metadata ontology, and governance architecture as system-of-record contracts for fraud, lending, and compliance.
• ▸Rebuilt executive trust after fraud-lending integration failure by tracing root cause to conflicting risk score definitions; introduced metadata registry, PII classification framework, and cross-BU governance working group.
• ▸Resolved VP-level escalation on real-time infrastructure by showing proposed architecture created 3x cost increase for <5% outcome improvement; defended hybrid model preserving detection performance while containing cost.
• ▸Operationalized fraud ML by standardizing feature pipelines and embedding model outputs into investigator workflows, eliminating manual triage across 5 systems, improving investigation throughput 29%, and reducing deployment time 30% across 40+ integrations.
• ▸Established ML deployment approval gates with Legal, Compliance, and fraud investigation teams, creating reusable governance standards for model readiness, PII handling, and operational adoption.

Owned product strategy for $270M enterprise API platform serving 200+ internal teams across consumer banking, commercial banking, and credit cards.

• ▸Drove 15% adoption growth and $80M+ measurable value creation by expanding platform capabilities, improving developer experience, and accelerating integrations across internal customers.
• ▸Built usage instrumentation tracking API consumption, error rates, integration bottlenecks, and downstream business impact; used signal data to prioritize roadmap investments toward highest-leverage platform capabilities.