Enterprise AI Solution Architect | Agentic AI Systems | Technical Program Leadership

Building production AI systems where technical reliability, enterprise governance, and measurable business value converge.

Beaverton, Oregon · SSOE Group · M.S. AI/ML, University of Michigan

Portfolio · LinkedIn · GitHub · Medium · Podcast · Email

I architect enterprise AI systems that operate in environments where reliability, security, and business consequences matter. My work spans semiconductor manufacturing, automotive production, and architecture, engineering, and construction (AEC).

At SSOE Group, I lead AI architecture and enterprise adoption initiatives across a 1,400-person engineering organization. I architected a production multi-agent RAG system supporting business development and proposal intelligence, led the enterprise rollout of Abacus AI, and help establish the technical and governance frameworks for building and deploying AI agents at scale.

My work connects three disciplines:

• AI systems architecture: Agent orchestration, retrieval-augmented generation, enterprise knowledge systems, deterministic workflows, evaluation, and observability.
• Enterprise AI transformation: Platform adoption, security governance, citizen development, vendor integration, and scalable deployment.
• Technical program leadership: Translating business requirements into architecture, coordinating cross-functional execution, and delivering measurable outcomes.

My background began in chemical engineering, progressed through semiconductor process engineering at Intel, and evolved into applied AI and enterprise architecture. At BMW of North America, I led an AI Champions initiative spanning manufacturing operations in the United States, Germany, and China.

I approach AI as an engineering discipline: models provide intelligence; architecture, validation, and governance make that intelligence dependable.

Production Architecture | SSOE Group

Architected a production multi-agent intelligence system serving more than 200 executives and business stakeholders.

• Specialized agents for market discovery, account intelligence, competitive analysis, sales enablement, and organizational knowledge.
• Intent-based routing, hybrid retrieval, and multi-source evidence synthesis.
• Enterprise identity and access control through Microsoft Entra ID.
• Architecture designed around reliable orchestration, persistent state, and controlled tool execution.
• Ongoing evaluation of managed agent infrastructure and SDK-based implementations for enterprise security and operational requirements.

Business value: Makes institutional knowledge and market intelligence accessible through a coordinated AI workflow rather than disconnected document searches.

1,400 Employees | SSOE Group

Led the enterprise rollout and governance of Abacus AI, enabling AI-assisted workflows across engineering and business functions.

• 80% adoption within 30 days, with enterprise deployment completed by day 90.
• Approximately 900 employee-built AI applications across business functions.
• Established an AI App Store governance model with a three-stage application review process.
• Defined usage controls, token budgets, application publishing standards, and adoption reporting.
• Coordinated enterprise AI platform strategy across Abacus AI, Microsoft Copilot Studio, and external AI vendors.

Business value: Turned enterprise AI from isolated experiments into a governed, broadly accessible organizational capability.

Azure | RAG Architecture and Research

Designed an architecture for complex engineering document collections where answers must be traceable to source evidence.

• Layout-aware, page-level indexing for engineering documents.
• Hybrid BM25 and semantic retrieval.
• Parallel multi-source retrieval and evidence reconciliation.
• Identity-aware information access.
• Evaluation of retrieval quality, citation fidelity, and system reliability.

Technical focus: Moving beyond basic chatbot RAG toward evidence-grounded enterprise knowledge systems.

Manufacturing AI | Delivered

Designed and led an AI enablement program across BMW manufacturing operations in Spartanburg, Germany, and China.

• Enabled plant teams to identify, prioritize, and develop manufacturing AI opportunities.
• Connected operational challenges with practical data science and AI capabilities.
• Built cross-site collaboration and knowledge-sharing practices.

Business value: Expanded organizational capacity to discover and execute applied AI initiatives closer to manufacturing operations.

Designing systems where LLMs operate within explicit execution boundaries rather than unrestricted autonomous loops.

Focus: LangGraph, Anthropic SDK, agent routing, state machines, checkpoints, bounded tool execution, structured outputs, human approval, and evaluation harnesses.

Building retrieval architectures that preserve source traceability, respect access permissions, and reconcile evidence across enterprise information systems.

Focus: Hybrid search, reranking, document parsing, citation grounding, Entra ID authorization, and retrieval evaluation.

Exploring how semantic models, business rules, and knowledge graphs can give AI agents a shared operational understanding of enterprise entities, relationships, and constraints.

Focus: Domain ontologies, knowledge graphs, policy-constrained decisions, deterministic rule engines, and auditable agent execution.

Establishing the infrastructure and controls required to operate AI systems across departments, users, and business processes.

Focus: Platform governance, application lifecycle management, IAM, observability, LLM cost controls, security reviews, and responsible deployment.

Connecting architecture decisions with enterprise outcomes through disciplined engineering execution.

Focus: Technical roadmaps, architecture tradeoffs, vendor evaluation, stakeholder alignment, operational KPIs, and cross-functional delivery.

I maintain independent engineering case studies to test architectures, quantify reliability, and document what works—and what does not.

These projects emphasize reproducible evaluations, real execution evidence, failure analysis, and explicit engineering tradeoffs.

Engineering principle: A system that fails a measurable evaluation is more informative than a polished demo that never defines success.

My career has moved from physical process engineering to intelligent software systems, but the underlying principle has not changed:

Complex systems must be measurable, controllable, and designed to fail safely.

The next generation of enterprise AI will require more than powerful language models. It will require architectures that combine agentic reasoning, organizational knowledge, deterministic business constraints, security, and reliable execution.

That is the intersection where I build.

Explore My Portfolio · View GitHub Projects · Connect on LinkedIn