Fuzzy problem → useful system.

I build systems that increase human capability. The recurring pattern: find a real-world ambiguity, understand the people and incentives in it, build the first useful system, validate it with evidence, and carry the reusable pieces into the next problem.

Santa Cruz, CA · Remote preferred · SF Bay Area / California hybrid for the right team.Santa Cruz, CA · Remote

Projects
projects

Selected work

Projects

Energy Scout

Makes solar ROI, incentives, and installer options clear for homeowners.

View project ↗

PropFi + Real Estate Ops

Connects property finance to a privacy-safe listing-to-leasing operating loop.

ABrain operating system

A private control plane connecting Claude Code, OpenAI Codex, Hermes Agent, and persistent specialist agents.

Bitcoin ML

Tests Bitcoin signals against forward returns with walk-forward and Monte Carlo validation.

View research ↗

How I work

Frame → build → orchestrate.

  1. FrameFind the decision the product should improve.
  2. BuildShip the smallest version someone can actually use.
  3. OrchestrateUse Claude, Codex, Hermes, Python, SQL, and integrations where each helps.
  4. VerifyKeep tests, evidence, approvals, and readable receipts around the system.

Multi-harness toolchain

Claude Code, OpenAI Codex, and Hermes Agent each serve different roles alongside Python, SQL, APIs, and task-specific integrations.

More about me

Background

Engineering → solar → data → applied AI.

Focus

Energy, property, markets, and tools that make decisions clearer.

Best fit

agent workflowsLLM appsforward-deployed AIworkflow toolsearly 0→1 systems

Good fit

ambiguous product/workflow problemsdata/integration-heavy toolsAI prototypes with real usersbusiness ↔ technical translationmission-aware teams

I'm strongest at 0→1 product and applied-AI work: I own discovery and the first working version, then hand off deep platform/infra hardening to specialist engineers.

Drag the branch or any card. Copy the link and send it back.