Native iOS and Android, engineered through agentic workflows.

Founding engineer on eBay Live, eBay's live shopping platform. Live video and real-time commerce, in the eBay app, on Android.

The work is the hard part of real-time mobile: ExoPlayer lifecycle and playback recovery, feed internals, keeping the commerce surface in sync with what is actually playing, and degrading gracefully when a phone drops off Wi-Fi mid-stream. The scale is eBay's. The failure modes are mobile's.

Alongside that, internal agent tooling and AI workflows for the live org.

I'm a Staff engineer with 13+ years of native mobile experience, currently a founding engineer on eBay Live. Over the last two years I've been rebuilding how I ship mobile software: native iOS and Android, developed primarily through AI coding agents with a human in the loop. This site documents that pipeline and the work it produces.

Shipping native iOS and Android apps where AI coding agents do most of the code production. This is ongoing practice, not a one-off experiment.

What's interesting is the scaffolding problem, not the models. Mobile has been slower than web to absorb coding agents: build and emulator loops produce thinner feedback than a web dev server, platform idioms are dense and version-fragmented, runtime state on a device is largely invisible to an agent writing code against it.

The engineer's role shifts. Less typing, more pipeline design. Structuring build output an agent can read. Piping runtime telemetry back into the loop so the agent can see what the running app actually did. Extending the agentic surface past code generation, into tests, stability checks, release. Curating prompt libraries so the agent isn't reinventing Compose idioms from scratch. Reviewing what Claude Code produces.

Thirteen years on Android means native idioms are where my instincts are deepest. The pipeline itself isn't framework-locked. It adapts to wherever the team lives.

Before eBay, I cofounded FreeMo in ad tech. Raised outside funding. Built the stack top to bottom: ad serving, auction logic, attribution.

The product was technically better than what beat it. Distribution decided it. That's why the apps on this page shipped to the stores early and kept iterating in public instead of waiting to be finished.

Golazo Services is the company: advisory, investments, and the AI-first mobile lab this site documents. Kognitiv is the label the on-device ML apps ship under.

Current focus: shipping apps through the pipeline, writing up what breaks, and helping mobile teams get real leverage from AI through the work itself, not through abstractions. The posts on this site are the documentation.

I'm looking for Staff or Principal work where real-time mobile and agent pipelines both matter. Fractional engagements run through Golazo Services, and I'm glad to compare notes with anyone building mobile coding agents.

Email me at [email protected]. The chat bar answers questions about my work; email's best for everything else.