The same move, four times
Three times I've built the layer an engineering org ended up standing on. OpenShift at UPS, and the org moved to microservices on it. Spring Boot at Spotify, then Productivity Engineering, where the job was making every other team faster. The internal agent platform at Airbnb: the Claude Code plugin, the MCP servers, the eval framework that gated what shipped.
Agents are the fourth, and they changed the job. A platform used to serve engineers. An agent platform serves anyone with work to hand off, and it acts instead of waiting to be called. So control stops being a preference and becomes the whole game: what an agent may touch, how you know it did the right thing, what happens when it didn't.
At Aphora Health I own that layer end to end, across engineering and operations. The product is people paying less for the medicine they already take, so the platform runs with PHI in the loop and no room for guessing.
This site is the argument
You're reading the output of the method. Posts come out of a voice pipeline I built. Images are generated. Prose gets linted before it ships. Publishing is scheduled and automated. A safety hook blocks agent commands I don't want run, and it has stopped me more than once.
It's all one public repo. The method is the repo.
Experience
Roles, employers, years. The rest is in the bio above.
Principal Engineer, AI Strategy & Agentic Platform · Aphora Health
2026 – presentOwn AI strategy and the agentic platform for the whole company, engineering and operations. Marketplace plugins, engineering platforms, automated security-vulnerability remediation, and agentic workflows for teams that don't write code. Pharmacy benefits, so PHI is in the loop.
Staff Software Engineer, Data Management · Airbnb
2024 – 2026Productionized an internal AI agent for natural-language search and discovery across the data warehouse. Claude Code Marketplace plugin combining skills, subagents, hooks, and commands. CLI + API MCP servers with internal AuthN/AuthZ. Evaluation framework that scores business outcomes, not unit pass rates.
Founder · TabletopAdventureCreator.comFounder track
2022 – presentSolo-built generative-AI SaaS for tabletop RPG adventure creation. In production since 2022, before the current generative AI wave.
Staff Software Engineer, Productivity Engineering · Spotify
2022 – 2024Chairman of the Productivity Engineering Technical Steering Group. Led six teams across developer tooling, IAM, and employee lifecycle. Reduced non-business-focused dev cycles by 30%.
Senior Software Engineer, Spotify for Artists · Spotify
2020 – 2022Two-sided marketplace work: GDPR audit system for artists, labels, and distributors. Payment provider scaling. Instructor for the internal data-science bootcamp.
Data Engineer · Spotify
2017 – 2020Introduced Spring Boot to Spotify's backend. GDPR compliance on GCP BigQuery. Co-founded a Google + Spotify Special Interest Group.
Associate → Senior Application Developer · UPS
2014 – 2017JVM platform work. Introduced OpenShift to enable microservices. Spring Cloud, AXON event sourcing, lambda-architecture streaming on Cassandra/Solr/Spark, JBoss Fuse / Camel / ActiveMQ.
B.S., Information Technology · NJIT
2009 – 2013New Jersey Institute of Technology.
Selected public work
Things you can open and check for yourself.
This site, as source
The blog and the pipeline that publishes it. Commands, safety hooks, prose linting, scheduling, and every post as raw MDX.
Tabletop Adventure Creator
Solo-built generative-AI SaaS for tabletop RPG adventure creation. Live since 2022, before the current AI wave.
@Mandalorian007 on GitHub
Public repos including aitk (the portable AI CLI toolkit), claude-code-toolkit, claude-tmux-manager, and other Claude Code tooling.
The Agentic Engineer blog
Field notes on handing real work to agents without losing control of your codebase. Patterns, tooling, things that broke.
Want me to come talk to your team?
Talks and workshops