A one-person consultancy built on original methodology research into AI-assisted software development.
I am a software architect with 25+ years of experience building production systems — from bioinformatics platforms and aerospace MRO enterprise software to hyperscale distributed systems at Walmart Labs. I hold a Ph.D. in Experimental Psychology from McGill University, which informs the empirical discipline I bring to methodology research.
At Walmart Labs I served as a Staff Software Engineer and architect on the rebuild of the company's multi-cloud push notification service — the backbone of the store pickup experience, required to process 30 million messages per day with 24/7 uptime and disaster recovery guarantees. Before that, at Tapestry Solutions and MIRO Technologies, I spent a decade on the architecture of GOLDesp, an aerospace MRO enterprise platform, gaining direct exposure to the traceability culture and reliability standards of regulated domains.
Marigold Intelligence LLC is a one-person practice. I work directly with engineering teams — no account managers, no junior staff, no subcontractors.
The AI coding era creates a specific problem that most teams are not equipped to see: AI generation makes structural decisions silently, at a speed that outpaces every governance model that depends on periodic review. Specification quality becomes the binding constraint that determines whether AI multiplies velocity or chaos.
I spent two years researching this problem systematically across three dimensions — development methodology, architecture governance, and verification — producing the three papers below. The consulting practice is the applied extension of that research: working directly with teams to close the gap between where their process is and where it needs to be.
The papers are available for review under a non-distribution agreement prior to publication. If you are an engineering leader evaluating this work, contact me.
Three connected papers currently under peer review. Together they form the intellectual foundation of the consulting practice.
AI shifts the binding constraint from implementation to specification. Neither Waterfall nor Agile ports cleanly into the AI era. The paper identifies the hybrid pattern best-performing teams are converging on: Agile cadence, Waterfall-quality specification, AI implementation.
AI makes structural decisions less visible, not less important. The paper proposes a continuous, specification-anchored governance model covering both regulated/safety-critical systems and commercial software with evolving requirements.
Introduces the SBR framework and its central concept: the trust boundary — the provability boundary of the test suite, defining the region within which autonomous AI generation is verifiable. The framework integrates EARS requirements, spec-driven methodology, architecture governance, TDD, and BDD into a directed chain for AI-assisted development.
The principles that guide every engagement.
Recommendations are evidence-based. If the SBR framework does not address a team's problem, I will say so. Clients deserve honest assessment, not a methodology sale.
The SBR framework is grounded in formal verification theory, empirical research, and practitioner evidence. The same intellectual discipline applies to every engagement.
I work directly alongside the engineers and engineering leaders I engage with — not as a detached advisor delivering slide decks, but as a practitioner embedded in the actual work.
Success is measured by what changes in how a team works — not by the number of sessions delivered. Every engagement is accountable to outcomes the team can see and verify.
This is a one-person practice with a defined methodology focus. I do not offer services I cannot deliver at depth. What I offer, I deliver personally and with full commitment.
AI-assisted development carries real consequences for software quality and team culture. The SBR framework keeps humans accountable for specification and design — by structure, not by hope.
Marigold Intelligence LLC is based in Vista, California — in San Diego County. Engagements are conducted remotely or on-site depending on the team's needs and location.