Software engineer. I build data platforms, cloud infrastructure, and tools that help teams and individuals get more from technology.
Serverless event-driven system at JPMC that coordinates multi-stage ingestion pipelines across the enterprise. I chose configuration-driven ETL over per-vendor custom code because every new integration was taking weeks of engineering time. Factory patterns now handle JSON, CSV, Parquet, XML, and XBRL from dozens of external vendors without code changes.
JPMC's internal platform for teams to search, evaluate, and access datasets across the enterprise. I managed the Kubernetes cluster running 8 containerized services and tuned Solr to handle the query volume when adoption grew faster than anyone expected.
End-to-end data pipeline for pharmaceutical image classification. Automated ingestion from NIH datasets with format conversion, validation, and metadata generation.
I can learn any tool. The value is what I enable with it.
If onboarding a new source requires a code change, the platform failed. I build systems that treat definitions as data.
Not something to optimize later. I reduced infrastructure costs by 70% by treating cost as a first-class requirement from day one.
Observability, runbooks, and clear failure modes are not optional. If the system can't explain what went wrong, it's not production-ready.
Then iterate. Dashboards delivered 6 months ahead of schedule because the first version was scoped to what actually mattered.
I pair with AI to survey a domain before making architecture decisions. Deep research in hours, not weeks. The judgment on what to act on is mine.
This site was designed and built in conversation with Claude Code. The interactions, layout, and content were iterated in real-time through rapid feedback loops.
AI generates. I decide. Every architecture choice, every design judgment, every “no, that's not right” is mine. The tool accelerates execution. The taste and judgment are not delegable.