I lead teams shipping agentic AI at enterprise scale.
13 years turning complex data into decision systems people actually use, from powertrain engineering to ML platforms to agentic AI. On my own time, I create public apps and run autonomous agents on real errands.
I spent eight years learning why vehicles fail, built a tool because finding root cause was too slow, and have spent the five years since turning that tool into an AI platform.
Six rotations across powertrain design, calibration, and vehicle launch. Even here, built tools to cut analysis time. The pattern starts.
Solved major warranty issues by hand, got frustrated with how slow root cause was, and built the tools to beat it.
Took the tool I designed zero-to-one as an enterprise ML platform, driving $100M+ in annual warranty savings.
Leading a 12-person team from ML to agentic AI: LLM pipelines, agentic summaries, and chat-with-data agents shipped, with multi-agent orchestration in development.
Annual warranty savings driven by the root cause platform taken from concept to enterprise adoption
Annual value delivered by the connected-vehicle AI platform I lead today
Served monthly by the platform's ML, LLM, and agentic capabilities
Earlier issue detection from LLM pipelines linking warranty claims to connected vehicle diagnostic data
2024 Henry Ford Technology Award, Ford's highest technical honor, for AI-powered root cause analysis
Three patents plus a trade secret for mobility and analytics innovation
Public web apps shipped with AI-assisted development, with one more in development
Internal automation tools personally built in the first 7 weeks of having OpenCode
Each build is live and public, and every app ships against the same checklist: revision history, tab icons, help pages, and security and privacy measures. Click through and use them, or check their uptime status.
Checkout-free retail computer vision running entirely in the browser: a CNN detects items by webcam and maintains a virtual cart, with debounce and presence logic as trust safeguards. No video leaves the device.
Resume advice and career guidance rooted in clarity and purpose, with an integrated Claude-powered chat coaching users in real time.
An interactive skill tree that turns "get better at AI" into a concrete, navigable path with tracked progress.
A mobile-first nutrition companion, installable as a progressive web app for quick everyday use.
An app that analyzes news articles, built toward a broader interest in media accuracy and reducing polarization through better information tools.
Not chatbot use. These are autonomous agents given access to my browser, files, calendar, and task manager, running real workflows with human oversight.
Creates my to-do list through the Todoist MCP, organizes it into projects and sections, then picks up the tasks it can handle itself and flags the rest with proposed next actions.
Checks WhatsApp for incoming messages and responds on my behalf, turning WhatsApp into a working channel that gets around Dispatch's limited chat interface, where links don't work. Senders are verified by phone number, not display name.
Pulls source files from Drive and produces deliverables directly back into it, so finished work is waiting in my Drive and viewable on the go.
Runs UX analyses of my apps with multiple personas, walking each through the product and producing a report of the user journey and pain points.
Dispatch agents exercise the web apps I build, validate flows, and produce structured findings reports.
Works from my shopping list, picks the items we need, and researches the best options to buy, then finds and tests coupon codes, builds carts, and runs buy/wait analyses before purchases.
Brief an agent with the full case file (serial numbers, purchase dates, issue history), let it converse with support agents on my behalf, and monitor periodically instead of sitting in the queue.
A maintained prompt library and production-grade autonomous task prompts with decision trees, security rules, and fallback behavior.
All work lives in projects, giving each its own custom prompts and file access, layered on a typed, indexed markdown memory that carries profile, preferences, and context across sessions, so no session starts cold.
2024 Henry Ford Technology Award, Ford's highest technical honor; 3 patents and a trade secret for mobility and analytics innovations.
Selected for Ford's inaugural Thirty Under 30 program; founded 2 professional organizations with 3,000+ combined members.
University of Michigan: M.S. Engineering, Energy Systems (2017); B.S. Mechanical Engineering, Math minor (2012).
LLMs & Generative AI, Agentic AI & Tool-Calling, Multi-Agent Orchestration, RAG, Machine Learning, Causal Inference
OpenCode, Claude (Cowork & Dispatch), Replit, GitHub Copilot, Prompt Engineering, MCP Integrations
Strategy & Roadmapping, Zero-to-One, Prioritization, KPI/OKR Design, Cross-Functional Leadership, Agile (SAFe), Design Thinking
SQL, GCP (BigQuery), Jira, Power BI, Tableau, R; Lean Six Sigma Black Belt