Ground the system
Give AI the right project context, knowledge, boundaries, and retrieval path.
Software Architect AI-Native Systems Engineer
I architect systems that make AI useful in production.
20+ years engineering production systems. I own architecture and delivery for financial platforms and applied AI.
Founder of RodGrid: I direct coding agents and review their work against the intended design.
- 20+
- years engineering production systems
- 1,000+
- architecture components in RodGrid
- 80%+
- ETL runtime reduction delivered
- 75%
- fewer Zapier tasks per transaction
01 / Point of view
Applied AI is a systems problem.
A model call is the easy part. The hard part is building a platform that knows its context, handles failure, controls cost, protects sensitive changes, and earns trust.
Design the failure path
Retries, idempotency, audit trails, fallbacks, and operator visibility are architecture.
Scale the review
Automate routine checks and increase human control as sensitivity and risk rise.
02 / Selected work
Systems, not slideware.
Architecture decisions made concrete: a collaborative AI platform, recoverable financial workflows, and measurable improvements in runtime and task consumption.
Independent productFounder
RodGrid
Architecture that carries through to implementation.
I designed RodGrid to carry system intent from collaborative modeling into AI analysis and engineering handoff.
AI coding agents produce most implementation code; I own architecture, product and technical direction, validation, and production acceptance.
- Realtime modeling with 1,000+ architecture components.
- Domain-aware AI agents with visible latency, token usage, cost, and failures.
- Architecture → specifications and implementation handoff, with code-vs-grid validation.
- Production boundaries across users, tenants, and grids, with reliability and release checks.
RodGrid engineering case studyArchitecture, AI orchestration & delivery
Architecture & scale
A React and PixiJS canvas models services, data flows, and operational boundaries using 1,000+ architecture components. Cloudflare Workers and Durable Objects support realtime collaboration and state coordination; D1 and R2 provide persistence and object storage.
The same model supports challenges, guided exercises, reference architectures, diagnostic and stress scenarios, and interview kits. Architecture validation, pattern detection, simulation, and SLO/cost analysis make the diagram something engineers can inspect and test.
AI orchestration
Domain-oriented analysts use task routing, role-specific prompts, and structured architecture context through the OpenAI Responses API. Explain and Analyze & Fix workflows, architecture evaluations, and AI scorecards build on that shared model.
AI execution is observable through per-agent latency, token usage, cost, and timeout/failure metrics. Local validation where appropriate and review of proposed changes keep the model response inside an engineering workflow.
Engineering handoff
Architecture intake and system modeling feed structured requirements, PRDs, specifications, and implementation context. Code-vs-grid comparison checks implementation against the intended architecture.
Reliability & governance
Realtime state requires deliberate Durable Object lifecycle management and isolation across users, tenants, and grids. Authentication, billing, AI credit controls, and operational visibility are part of the platform design.
Prompt-injection classification and review, sanitization, automated tests, and browser verification support security and release decisions. Sensitive paths receive additional validation before production acceptance.
AI-native delivery
I define architecture, features, requirements, constraints, and acceptance criteria. Coding agents perform the majority of implementation and iteration using persistent project context. I review designs and output, validate behavior, make security and reliability tradeoffs, and decide whether the implementation meets the intended design.
Reliable financial integrations
Problem. Internal finance workflows had to coordinate with external accounting systems and recover from provider failures.
Engineering. Designed an Azure-native integration service for Xero, QuickBooks, and Dynamics 365 Business Central, with provider mappings, OAuth, webhooks, queues, and durable state.
Result. Idempotent processing, reconciliation, end-to-end traceability, and explicit failure recovery across asynchronous financial workflows.
Rules, documents & human review
Problem. Compliance rules and document workflows needed to evolve beyond legacy rule-engine behavior.
Engineering. Worked on a dedicated .NET Compliance Service with Cosmos DB and event-driven processing, plus OCR and document-AI POCs using OpenAI, Gemini, and Azure Form Recognizer.
Result. Document analysis connected to compliance workflows, with human validation and controlled handling of sensitive information.
Less time running ETL
Problem. Long-running SQL Server ETL pipelines at Remsoft.
Engineering. Reworked query architecture, indexing, and stored procedures.
Result. Reduced execution time by more than 80%.
Fewer Zapier tasks
Problem. The initial integration consumed approximately 12 Zapier tasks per transaction.
Engineering. Tested Code by Zapier to consolidate actions, then built a custom Node.js integration with Zapier CLI.
Result. 12 → 3 tasks per transaction: a 75% reduction in task consumption.
Project knowledge in the workflow
Problem. AI-assisted development needs relevant project rules and architecture context.
Engineering. Built a portable Node.js and Qdrant MCP context server to index that knowledge, alongside Cosmos DB and GPT query-analysis utilities.
Result. Retrievable project context and practical analysis tools within the engineering workflow.
Recovery and visibility in production
Problem. Queue failures, batch processing, and slow shared services needed operational attention.
Engineering. Improved retry automation, Service Bus processing, and TaskRunner performance; contributed to the move from file-based logs to Rollbar.
Result. Better recovery and error visibility, backed by production debugging and follow-through.
03 / Expertise
What I bring to the room.
From financial integration and distributed platforms to applied AI and technical leadership.
Applied AI & LLM systems
Document intelligence, RAG, and domain-aware AI orchestration. Structured context, model-change evaluation, observable execution, and human review make outputs usable in real workflows.
- OpenAI
- Claude
- Gemini
- Azure AI Search
- Qdrant
Financial & compliance integration
Secure, asynchronous integrations across internal systems and financial platforms, designed around reconciliation and recoverability.
- Xero
- QuickBooks
- Dynamics 365 BC
- OAuth 2.0
- Webhooks
Distributed platforms
Backend and event-driven architecture with explicit service boundaries, durable state, queues, failure recovery, and observability. Deep C#/.NET experience from legacy modernization to cloud services.
- Azure
- .NET
- Service Bus
- Cosmos DB
- Docker
AI-native engineering
Direct coding agents through specifications and persistent project context; use design review, automated checks, and browser verification to accept their work.
- Coding agents
- Persistent context
- Automated verification
- Human control
Technical leadership & architecture
Turn ambiguous business needs into system boundaries and delivery plans. Lead design discussions, explain tradeoffs to stakeholders, coordinate developers and business analysts, review implementation, and mentor engineers.
- Technical direction
- Design & code review
- Mentoring
- Production ownership
04 / Experience
Built across the stack. Led across the lifecycle.
Production experience across banking, HR and staffing, startups, forestry, education, and R&D — with recent work focused on financial integrations, compliance, and applied AI.
Founder
Created RodGrid, an independent platform for system design and engineering handoff. I lead product and technical direction; coding agents produce most implementation under my review.
Explore the engineering case study for architecture, orchestration, and delivery practices.
Senior Software Engineer
Solution Architecture (functional scope)
Designed an Azure integration platform connecting internal financial systems with Xero, QuickBooks, and Dynamics 365 Business Central. Combined Functions, Logic Apps, Cosmos DB, and Service Bus with Entra ID/OAuth, webhooks, and recoverable asynchronous workflows. Also contributed to secure file workflows and SFTP integration guidance.
Evolved legacy rule-engine behavior toward a dedicated Compliance Service and worked on document intelligence using OpenAI, Gemini, and Azure Form Recognizer. Drove POCs and technical reviews for document/OCR workflows involving PII, with human validation and controlled reveal of sensitive information.
Improved queue retries, Service Bus processing, batch behavior, shared-service performance, and error monitoring. Provided architecture guidance and code reviews, coordinated delivery across developers and business analysts, and followed issues through production.
Senior Application Developer / Data Analyst
Reduced ETL execution time by more than 80% through SQL indexing, query redesign, and stored-procedure architecture improvements; built geospatial and operational tooling with FME, Python, React, Node.js, MongoDB, and AWS.
Senior Software Developer
Re-architected web applications spanning a Linux-based device and AWS, improving device-to-cloud communication, interface quality, and user management.
Application Developer
Modernized legacy .NET MVC and AngularJS applications and developed secure, API-driven recruitment and communication systems from the ground up. Implemented token-based authentication and external identity integrations using OAuth 2.0, OWIN, ASP.NET Identity, and Web API.
IT Project Leader / Senior Web Developer / Consultant
Led technical delivery and mentored junior team members across client projects, including work for Tecnisa. Designed SAP and factory-machine integration systems with event notifications and operational monitoring, and modernized Votorantim web and payment-calculation systems using .NET, SQL Server, WCF, and web technologies.
Earlier career
Senior development roles including Bradesco, J.P. Morgan/Vastera, and public-sector systems, focused on large-scale applications, SQL Server, ASP.NET, and enterprise integrations.
05 / Technology
A broad toolkit, used selectively.
- AI & data
- OpenAI · Claude · Gemini · Azure AI Search · Form Recognizer · LangChain · Qdrant · Pinecone · RAG · embeddings · Cosmos DB · SQL Server
- Cloud & integration
- Azure Functions · Logic Apps · Service Bus · Blob Storage · Azure SQL MI · Cloudflare Workers / Durable Objects / D1 / R2 · AWS · Docker · Kubernetes · OAuth 2.0 · webhooks · REST · SOAP · SFTP
- Engineering
- C# · .NET Framework/Core/8 · ASP.NET · Entity Framework · TypeScript · Node.js · JavaScript · React · Angular · SQL · Python for scripting and data workflows
06 / Role fit
Where I do my best work.
Complex platforms where architecture, applied AI, and distributed systems meet — especially when the problem is still ambiguous.
Staff, Lead, and Architect roles are a strong fit when they combine technical direction with responsibility for delivery and operation.
07 / Contact
Let’s talk about the role.
Based in New Brunswick, Canada. Let’s discuss your team, its backend or AI platform challenges, and the technical responsibility the role needs.