Austin, Texas Metropolitan Area
AI systems architect and full-stack engineer. I build AI products that ship safely and stay up. Background spans hardware validation (Texas Instruments), production software (Microsoft, Toyota/Pariveda), and independent AI product development — 10+ years with a consistent throughline: automate what should be automated, ship what needs to ship, and don't cut corners on the stuff that breaks at 2am. At Microsoft, I was the senior frontend engineer and frontend lead on my team, with pre-release internal access to GPT-3 before ChatGPT existed. I prototyped one of the earliest embeddings-based AI features in the org. Now I'm building RAIN — a relationship-first AI collaborator platform. 43 custom AI personas, 44 real users, ~4,000 conversations. Not a demo. Not a pitch deck. A live system built on trust-preserving AI patterns where the model is designed to disagree with you, not just agree. Under the hood: LangChain, LangGraph, FastAPI, Next.js 15, TypeScript, Postgres, ChromaDB, HyDE retrieval, Terraform, Azure, GitHub Actions CI/CD. 90%+ test coverage. Zero regressions. Every deployment gated by automated test suites. I also work with solo founders and small AI teams who shipped fast with Claude or GPT and now need senior eyes on what's actually production-ready. Architecture risk maps, secret and exposure audits, CI/CD reviews, and prioritized remediation — clarity before launch, not cleanup after disaster. If you're hiring for senior AI engineering or need a second set of eyes before you ship, message me.
I founded RAIN to build relationship-first AI collaborators — individualization over mass appeal. The leverage isn't in the model. It's in how the model is constrained, challenged, and allowed to disagree with you. - Built 43 custom GPT personas serving 38 users across ~4,000 conversations via OpenAI GPT Builder - Designed trust-preserving AI patterns: constraint-based personas built to disagree, challenge, and preserve operator judgment — not yes-man outputs - Building Riven: a persistent AI co-pilot with HyDE retrieval, ChromaDB vector store, and multi-dimensional memory scoring — 90%+ test coverage, zero regressions, deploying to Telegram beta - Built an agentic lead generation system with automated cron pipelines across LinkedIn, Reddit, and developer communities using OpenClaw - Architected and deployed end-to-end AI projects on Azure (Container Instances, App Service, Container Apps, Service Bus, Blob Storage) to validate a reusable IaC deployment toolkit - Built the Artefex Registry: Terraform modules, GitHub Actions CI/CD, ACR image management — repeatable infrastructure for rapid AI product launches - Full-stack AI systems using LangChain, LangGraph, FastAPI, Next.js 15, TypeScript, Postgres - TDD discipline throughout — Pytest, Jest, Playwright; every deployment gated by automated test suites What I've seen firsthand: users breaking out of paralysis, closing learning loops, and regaining momentum — when the AI wasn't a yes-man. One user used their AI as part of a broader support system during addiction recovery. Another used it as a reflective mirror that amplified their own intuition at speed. The goal isn't to cast a wide net. It's to help each person build the right relationship with the tool — because when that's right, everything else accelerates.
- Senior frontend engineer and frontend lead; owned end-to-end feature delivery across the product with strong QA discipline - Prototyped an embeddings-based Insurance Chatbot on GPT-3 with pre-release internal access — among the earliest hands-on LLM work at Microsoft, influencing the team's AI feature direction - Built a reusable wizard UI pattern adopted across the product, streamlining complex data entry flows and accelerating future feature launches - Mentored new engineers and filled leadership gaps post-layoffs; expanded ownership across frontend, backend, and project management - Drove adoption of conversational UI patterns and LLM integration strategy for the product roadmap Actively promoted mental health awareness, contributing to a more inclusive and supportive work environment
- Frontend architect on Project Windcrest (caregiver web app) — React, Apollo GraphQL, Docker, Azure, TypeScript - Built the GraphQL query layer, authentication service (MSAL), and feedback integration system - Led frontend through successful Alpha release; recognized for architecture quality and design system contributions - Shared expertise through pair programming and PR reviews, contributing to team knowledge and efficiency
– Improving developer efficiency, enabling more complex application features and reducing overhead of cross-training developers by spearheading a migration of Xamarin application to React Native – Increasing adoption of strong patterns, reducing use of antipatterns and training team on best practices as team quality champion with PR approval responsibilities – Reducing bug throughput, enabling code refactoring and providing visualizations of code coverage by implementing automated testing with Jest and test coverage reports in Jenkins pipeline using Cobertura – Providing team with framework to avoid common Javascript bugs by implementing conversion to Typescript across team repos – Improving overall success of team through mentorship of new team members
– Developed new features and improved code quality on Engage mobile application using Xamarin Native in C#, XCode Interface Builder and XAML – Implemented shared functionality between applications using AWS Lambda backend – Developed new features and improved code quality on Engage React web application – Improved deployment process using Jenkins CICD pipeline
– Uncovered defects and enabled characterization and final testing by designing and executing validation plans for analog front ends on battery packs in electric vehicles – Revolutionized validation process by developing a set of standardized libraries for automating lab bench equipment using Python, LabView and TestStand and designing a relay matrix testing ecosystem to automate test station reconfiguration – Provided team insight into device characteristics by analyzing data and creating visualizations using Python scripting and Spotfire – Stressed silicon and found design issues by creating “break the part” style tests and spread lessons learned across TI in presentation at Verification and Validation Conference