Financial Engineering · AI-Native Data Systems · Event-Driven Research
I build research infrastructure for turning messy public information—filings, central-bank communications, market data, and technical documents—into structured signals, dashboards, and decision-support systems.
Selected Work
Claude Code backend/model benchmark on Terminal-Bench/Harbor with adjusted cost accounting, pass-rate frontier analysis, and nonproductive spend tracking.
LLM-assisted pipeline for acquiring Federal Reserve communications, generating structured monetary-policy sentiment scores, and visualizing sparse, irregular central-bank communication signals for portfolio research.
Local-first SEC filing infrastructure for turning filings, exhibits, and event disclosures into reproducible research artifacts and structured event signals.
Prototype architecture for agent-assisted investment research: ingest documents, extract structured events, generate trade hypotheses, run backtests, and write experiment postmortems.
Fan-augmented ground-effect vehicle concept designed as a physical platform for validating controllable downforce, active chassis control, and force-aware vehicle-system IP.
Builder Background
My work sits at the intersection of financial engineering, software systems, and physical prototyping. Before focusing on AI-native finance and data infrastructure, I co-founded Opentrons Labworks and helped build early liquid-handling robotics systems, including firmware and Python API work for the OT-1.
That builder background now informs my approach to research systems: make the pipeline observable, reproducible, and useful enough that others can test and extend it.
Published Materials
Canonical web resume with cleaned public-facing content.
A sponsor-facing technical report covering Federal Reserve communication scoring, FOMC ingestion, OpenAI Batch API processing, score materialization, visualization, limitations, and continuation recommendations.
Contact