Work with me
Consulting and freelance builds — AI systems, automation, and software that quietly does its job.
What I build
- Agentic systems & LLM tool-use — assistants that do things rather than chat about them: tool surfaces (MCP), multi-model routing, cost guardrails, memory that holds up.
- Serverless backends — AWS Lambda, DynamoDB, CloudFront, Terraform. Small, legible, least-privilege systems that don't need a pager.
- Native apps with AI inside — iOS apps backed by stateless, privacy-preserving proxies: bring-your-own-key, per-device cost caps, nothing retained server-side.
- Automation & pipelines — the unglamorous glue: scheduled jobs, notifications, capture, reporting.
Evidence
A personal AI agent, in production
An SMS-and-voice assistant on AWS in daily use: ninety-odd single-purpose Lambda functions behind one authenticated tool surface, scheduled automations, event-sourced memory, per-function least-privilege IAM — and it builds and deploys much of itself.
python · aws lambda · dynamodb · terraform · mcpThis website
Self-modifying by design: pages are data the system writes at runtime — the agent added this page without a code deploy. Lambda-rendered behind CloudFront; the entire design system is one stylesheet.
cloudfront · lambda · s3 · one css fileNative apps, private AI
iOS apps whose AI features run through stateless proxies — keys and content stay the user's; the server keeps nothing. Shipped through cloud-signed CI with no Mac in the loop.
swift · swiftui · github actions · testflightHow engagements work
Short advisory engagements or fixed-scope builds, remote. Tell me the problem and what done looks like; I'll tell you honestly whether it's a week, a month, or a bad idea.