Annual interactions
Designed AI workflow architecture for Walmart's enterprise care platform, including LLM interpretation and deterministic policy execution.
Product builder · iOS tinkerer · AI workflow maker
I am Matthew Coelho, a product builder with 11+ years in software and AI-driven workflow systems. I have shipped enterprise AI at Walmart scale, solo-built iOS apps, and turned agent ideas into working demos people can actually try.
Experience
The side projects are hands-on, but the operating background is enterprise scale: AI workflow architecture, human-in-the-loop systems, evaluation design, and platform adoption across complex teams.
Designed AI workflow architecture for Walmart's enterprise care platform, including LLM interpretation and deterministic policy execution.
Built workflow infrastructure and guided tools used by operators across e-commerce, grocery, pharmacy, marketplace, and delivery.
Shipped AI co-pilot capabilities that reduced operator task completion time at launch, with rework and trace data as guardrails.
Solo-built iOS apps, AI workflow tools, agent demos, and small browser games that people can open, test, and use.
Selected work
A mix of shipped apps, AI prototypes, and experiments. Each project is chosen because it shows a different muscle: product thinking, interface craft, automation logic, or fast learning.
Public site for the Diogo app family, supporting a solo-built iOS product live across 17 countries plus two App Store kids apps.
Open projectAgentic recipe assistant with 724 recipes indexed in Supabase pgvector and Gemini function calling across search, ranking, counts, and comparisons.
Open projectPlain-English workflow builder that turns process descriptions into structured visual workflows with nested branches, simulation logs, and n8n export paths.
Open projectState-machine conversational AI demo where guardrails gate progress behind verified tool execution rather than model confidence alone.
Open projectA playful browser game built as a quick toddler distraction, with five nursery songs and bright keyboard feedback that is easy to understand.
Open projectApp Store
The Team Diogo app family includes a social photo app plus two playful apps designed for children and families.
Anonymous nearby photos, swipe voting, comments, sharing, regional unlocks, vote-weighted ranking, and push notifications.
View App StoreA children-friendly ocean app with local, private gameplay and sea-themed characters.
View App StoreA simple, bright iOS game experience built around approachable play.
View App StoreAbout
I like products that can be touched quickly: an app, a prototype, a running demo, or a workflow that makes the idea concrete. The through-line is practical iteration, clear user value, and enough craft that the thing feels real.
At enterprise scale, that means deciding what the model owns, what deterministic systems must enforce, and where humans stay in command. In personal projects, it means shipping the smallest working version and learning from the real thing.
Toolkit
The projects here pull from native iOS, React, Python, Streamlit, Flask, Supabase, LangChain, LangGraph, Groq, Gemini, and static site deployment. The platform work adds eval design, trace analysis, deterministic policy execution, and human-in-the-loop workflows.
Contact
For product conversations, AI workflow ideas, or collaboration, LinkedIn is the cleanest place to start.