16+ years building and scaling software platforms in retail and e-commerce — from an in-house voice AI product used in thousands of stores, to the MCP-based agentic AI platform and edge-and-cloud infrastructure running underneath it.
I lead engineering for retail mobile, digital, and applied AI platforms at Tractor Supply Company, where I own strategy and delivery across a 25+ person organization spanning mobile, platform engineering, and AI products.
My work tends to start with a build-vs-buy decision and end with something running at scale: an in-house voice AI platform now used across 2,400+ stores, a computer vision system reading live camera feeds in 1,000+ locations, and an MCP-based agentic AI platform that lets both, and everything since, share the same infrastructure instead of becoming siloed one-off projects.
I care as much about what holds a platform up — observability, FinOps, identity, reliability — as I do about what ships on top of it.
An in-house, voice-based store communications platform built from the ground up on Sensory AI models for intent detection — replacing a third-party vendor and becoming the first product Tractor Supply built and operated itself rather than buying.
Took it from a build-vs-buy decision through a phased rollout to 2,400+ stores, adding intelligent notifications and real-time operational insights that help team members serve customers faster on the floor.
A store camera and AI system reading live feeds for traffic and dwell insights — built on YOLO models, and validated with a controlled 60-store study before wider rollout.
The pilot showed a statistically significant lift in conversion, and follow-up coaching missions tied to its alerts drove measurable improvement across most participating stores within the first week.
TSC Connect, Tractor Vision, and a set of store AI agents all run on the same infrastructure — Kubernetes-based edge servers in every store, paired with a cloud backend on Azure for model management, analytics, and integration.
Rather than three siloed AI projects each reinventing infrastructure, this turned them into one connected platform other teams could build on.
An MCP-based internal platform with reusable services and shared APIs, letting product teams build and scale their own AI applications without duplicating infrastructure.
It also powers customer-facing work like Ask Scout, Tractor Supply's conversational product discovery experience — I partnered with e-commerce engineering to build the OpenAI- and MCP-based services behind its product search, comparison, and cart actions. Its first internal use was a RAG-based HR chatbot that cut support-center call volume by 32%.
A super app for store team members, hosting a growing set of micro apps behind a single sign-on instead of a dozen separate logins.
Timekeeping, training, certifications and licensing, and other team member productivity tools all live inside it as individual micro apps rather than standalone systems.
Led migration of the website and mobile apps from on-prem infrastructure to Azure Kubernetes Service, with canary-based deployments that catch failing releases early.
Ran the FinOps side personally — right-sizing workloads and building capacity forecasts so the move came with a lower, more predictable cloud bill.