IT & security engineer relocating to Austin, TX or the Washington, DC metro. I like being the person who owns the whole stack: identity, devices, security, the network and the SaaS sprawl in between. I automate the routine work, lock down the rest, and keep it simple enough that a small team can run it. Available within a month.
I build and run IT for fast-growing companies. As the first IT hire at a software scale-up, I built the IT and security function from the ground up while headcount grew from 60 to over 300. Today I run identity, endpoints, security and collaboration for a 300-person fintech. Master's degree, CISM, Security+ and AWS Solutions Architect. Outside work I built and launched TensorFit, an AI fitness app, and run a self-built RTX 5090 home lab for OpenClaw agents and local models.


A cross-platform AI fitness coach, built solo and launched end-to-end on React Native / TypeScript with a fully serverless AWS backend (Lambda, API Gateway, S3) defined as Infrastructure-as-Code, and live on the App Store and Google Play.
Most of what I automate is the routine identity and device work: onboarding, offboarding and access changes on n8n and Microsoft Graph. These are the AI tools I run or look after.
I ran a pilot of Microsoft Scout for internal IT, with the agents running under Entra-governed identities so they get the same access controls as a person. The aim is to take routine requests off the queue so nobody has to work through them by hand.
I review AI tool requests against the acceptable-use policy and manage the AI licences and seats. Netskope shows which AI sites people are actually using, and its web filtering and DLP policies enforce which ones are sanctioned. CrowdStrike's AI detection covers what is running on the endpoints.
The data team wanted their Databricks reports searchable in Copilot. I set up a Microsoft Graph integration that writes the finished reports into SharePoint, with the app limited to that one site. Copilot indexes them and people can ask questions instead of chasing files.
I built an RTX 5090 workstation so my agents run at home. OpenClaw handles the always-on assistant and automation work, Claude Code handles coding, and DeepSeek Harness is the open-source agent runtime I run locally. Everything defaults to a local Qwen 3.8 27B through Ollama, and when a task actually needs a frontier model it switches to Claude Fable 5.1.