Advisory

I work with insurtechs, enterprise carriers, and AI-forward companies on technology questions that need senior judgment more than incremental hands. Engagements typically center on three areas: making AI usefully part of the software development lifecycle without breaking what works, modernizing legacy systems without stalling the business, and standing up engineering operating models that scale across geographies and regulatory regimes.

Engagement structures

Advisory retainer

Ongoing strategic partnership for founders and executives who need a senior technology partner in the room — for board discussions, hiring decisions, architecture reviews, and roadmap setting. Typically one to two days per month over a multi-quarter horizon.

Embedded engagement

Deeper involvement for time-bound initiatives: modernization programs, M&A integration, AI-SDLC rollouts, or fractional CTO coverage during a leadership transition. Typically several days per week over a three-to-nine-month engagement.

Project-based work

Defined scope with a defined deliverable: architecture reviews, technology due diligence, evaluation framework design, or focused workshops. Suited to questions that need senior expertise applied to a specific decision rather than ongoing partnership.

Domain focus

Insurance and insurtech

Fifteen years inside a Fortune 100 carrier across US and Asia-Pacific markets, with direct experience in personal and commercial lines, D2C platforms, distribution-partner integration, claims, underwriting decisioning, and M&A integration. The unusual span is not just the scale (200-engineer organizations, USD 24M budgets) but the regulatory and market diversity — a regional remit covering multiple jurisdictions, each with its own supervisory regime.

AI in the software development lifecycle

Led the design and rollout of AI-enabled SDLC and engineering workflows across a multi-market regulated enterprise, including the work to define how teams move AI use from R&D through pilot to scaled embedding in mainstream development. This included guardrails for quality, security, and model risk; productivity measurement that resists gaming; and patterns for AI-assisted legacy modernization (decomposing stored procedures, refactoring data models, generating documentation for undocumented systems).

AI evaluation methodology

Current applied research focus. Building an evaluation platform for a specific class of LLM failure mode — silent source-blending, where models synthesize authoritative sources that have actually split on an issue and present the result as consensus. The methodology generalizes across any domain with authoritative-source disputes (legal circuit splits, medical guidelines, regulatory standards across regimes).

If you'd like to discuss whether an engagement makes sense, email me or find me on LinkedIn.