Production Product AI
Led delivery of Root Cause capability for GHX Fields and E-Luminate at Syngenta North America, using AI to proactively analyze agronomic and environmental patterns and surface concerns plus mitigations for crop fields.
AI in production and engineering practice
AI is useful when it reaches production.
My AI work is grounded in enterprise software delivery: real product capabilities, report generation, AI-assisted engineering workflows, and local model investigation. The through-line is practical value, reviewability, and technical judgment.
Work streams
Four ways AI shows up in my work.
AI-Powered Reports
Built AI-assisted report-generation workflows for production client releases where narrative output needed to be grounded in operational data and reviewable business context.
AI-Assisted Engineering
Introduced pragmatic AI-assisted development practices at Skyward, especially for senior developers working with large systems, code review expectations, and repeatable delivery workflows.
Local Model Investigation
Explored hosted versus local models using tools such as Ollama and vLLM, with attention to privacy, latency, cost, workflow ergonomics, and operational fit.
Production AI example
Root Cause inside a Global Fortune 500 platform context.
The value was not a chatbot bolted onto a product. It was AI analysis embedded into field and agronomic workflows, where users needed earlier visibility into patterns, concerns, and possible mitigations.
Operating principles
How I think about AI adoption.
- AI should sharpen decisions, not hide uncertainty.
- Generated output needs review paths, context boundaries, and observable failure modes.
- Developer AI is most valuable when it improves flow without weakening engineering judgment.
- Hosted and local models have different privacy, cost, and operations tradeoffs.
Open-source tooling
Small tools that show the shape of the work.
These are not meant to be a second resume. They show how I think about AI workflow surfaces, delegation, local models, and practical developer tooling.
task-board
AI Kanban harness for managing agentic work, integrating with your existing workflow and tools.
Product thinking for AI control surfacescodex-image-gen
Portable Node CLI and skill wrapper around Codex image-generation workflows.
Practical AI toolchain integrationlocal-llm
Local LLM investigation space for model hosting and workflow experiments.
Hosted vs local AI tradeoff explorationpopcorn
REST API middleware for response shaping, reducing endpoint sprawl while giving clients precise payload control.
API/platform engineering depth