About
A bit about me
I'm a Software Engineer at PMG in Dallas, where I build AI tools that automate media workflows for multi-million dollar ad campaigns. My work has directly improved ROAS and saved hundreds of hours across teams—stuff that actually moves the needle.
What I enjoy most is taking an idea all the way to production. Not just writing code, but thinking through the architecture, setting up proper monitoring, and making sure it runs reliably. I like understanding why a system is structured a certain way, not just how it works.
My go-to stack is Go and Python on the backend, React on the frontend, and PostgreSQL for data. I'm AWS Cloud Practitioner certified and genuinely enjoy the infrastructure side—Datadog, New Relic, Docker, the whole observability stack.
Outside of work, I'm usually playing volleyball, exploring Dallas on foot with my walking group, or working on side projects. I'm always building something.
Experience
Where I've worked
- Engineered a multi-agent workflow automation platform in Go with ADK that converts natural language queries into fully configured, executable workflow automations, saving 5 hours/week for 1,200 users. Boosted semantic search accuracy by 20% via cosine distance evaluation across top-k vector embeddings augmented with vendor-specific keyword search
- Developed a Slack-integrated RAG chatbot using Python, LangChain, and the Confluence API to provide client-team-specific onboarding support in real time, saving 150+ hours/week across 50 media coordinators
- Built a predictive forecasting tool using weather data, statistical regression, and LangChain—delivered ~5% lift in ROAS for major QSR clients managing multi-million dollar budgets across 5 media teams
- Built a Python application integrating with Google Ads Platform API to generate campaign bulk sheets, reducing setup time by 3–5 hours per campaign launch and scaling across 25+ clients
- Designed a domain-negation application in Python that programmatically excludes unsafe/low-value sites via Bing Ads API across 15+ search accounts, increasing ROAS by 10%
- Architected an offline conversions uploader in Python for Google & Bing Ads to enhance Data Augmented Bidding for Search Ads, processing 400,000 conversions daily across Intuit TurboTax & Buildertrend
- Eliminated $12,000+/year in cloud costs by replacing a legacy platform with a scalable API-driven architecture
- Mentored 5 junior engineers on Python best practices, observability, and Agile methodology
- Built an LLM-driven automation system using GPT-4ALL to generate and deploy audited system configurations for Disney's cloud infrastructure
- Achieved 35% reduction in system configuration management time; presented findings directly to Disney stakeholders
- Developed a full-stack parking management system using React & Express, serving 1,000+ users
- Designed the relational database schema and built a complete suite of RESTful API endpoints with validation
Projects
Things I've built
Ark
FeaturedHomelab asset tracking platform I built solo, end-to-end. Go backend, React frontend, PostgreSQL, deployed on AWS.
- RESTful API with clean architecture — handlers → services → repositories
- Rate limiting middleware for API protection
- PostgreSQL 16 with GIN indexes and tsvector for full-text search (<100ms queries)
- Redis for background job queues (Asynq)
- Database migrations with Tern for version-controlled schema evolution
- JWT middleware with two-phase auth — Clerk SDK token verification + user-scoped authorization
- Multi-tenant data isolation — all queries scoped by user_id
- Zod schemas → OpenAPI spec → generated TypeScript client
- Compile-time contract enforcement between React frontend and Go backend
- Monorepo with shared
packages/zodandpackages/openapi
- New Relic APM with distributed tracing
- Request-scoped structured logging with trace ID correlation
- Health endpoints validating database and Redis connectivity
- AWS EC2 (t3.micro) with Elastic IP and automated EBS volume lifecycle
- Docker Compose orchestration with Caddy reverse proxy (auto-HTTPS)
- GitHub Actions CI/CD — build, test, push to GHCR, deploy on merge
- Saving media coordinators 10+ hours/week on asset tracking
- RAG pipeline with pgvector embeddings and OpenAI
- Semantic search via AWS Lambda (cost optimization)
Book Forum
Weekend build to explore Firebase's real-time sync and Gemini's recommendation capabilities—different constraints than my usual Go/PostgreSQL stack.
- Gemini API for personalized reading suggestions based on discussions
- Firebase for real-time sync, auth, and storage
- Full CRUD for posts, comments, likes, and reading lists
Stack