Forward-deployed engineer for early applied AI work.
I write software, but I do not start with code. I start with the system: the workflow, the people, the data, and the decision the product needs to support.
The short story: I came to AI through real problems in solar, markets, property, and self-directed building — then kept building tools around the parts that felt unclear.
Formal resume
Alex Horton
horton.alexander@gmail.com
alexhorton.dev/contact
Profile
Forward-deployed engineer focused on applied AI, workflow tools, data products, and early product systems. Strongest where the product shape is still forming: mapping the workflow, building the first useful version, connecting APIs/data/UI/LLM tooling, and iterating with users or product-minded teams. Ten years of self-directed work across solar, markets, and physical systems gives me a grounded sense of which problems are worth solving and what "working" actually means.
Skills
Full-stack: Next.js, React, TypeScript, Tailwind, Node, FastAPI-style APIs, Python, REST APIs, PostgreSQL/Supabase, DuckDB, auth, schema design, scheduled data pipelines, Vercel, Docker, GitHub Actions.
AI/data: LLM workflows, Claude Code/Cursor/Hermes, tool-calling loops, OpenRouter/model routing, RAG patterns, persistent context, Pandas, NumPy, scikit-learn, XGBoost, feature engineering, backtests, evaluation loops.
Product: MVP scoping, user interviews, workflow mapping, dashboards, data visualization, ROI modeling, product analytics, technical translation.
Domain: Financial markets & crypto, solar/energy economics, property analytics, sovereign finance, market timing & risk frameworks.
Experience
- Bitcoin ML — Built a public quant research site around Bitcoin price drivers: Python data pipelines, feature tests, walk-forward backtests, reproducible dashboards.
- Energy Scout — Built a consumer solar calculator combining production assumptions, incentives, quote data, and ROI into a personalized report for homeowners evaluating solar.
- Agentic workflows — Designed and operate a coordinated AI development/workflow system with Claude Code, Hermes-style agents, model routing, persistent context, scheduled jobs, Telegram, and voice summaries.
- Self-directed equity & crypto research and trading — the direct source of the market, energy, and sovereign-finance domain fluency behind the products I ship.
- Designed, built, and maintained a solar-powered camper van (electrical, plumbing, carpentry) and lived and worked from it for years — real-world systems engineering under real constraints.
- Ran hands-on sustainability systems (aquaponics) — reinforced the physical-systems and product instincts I bring to software.
- Studied public markets, solar/energy infrastructure, Bitcoin, and macro systems to understand how capital, operations, and technology interact.
- Analyzed multi-source datasets (utility rates, incentives, consumption patterns) to identify underserved market segments in California multifamily solar and agricultural sectors.
- Built ROI projection models and presented data-driven recommendations to stakeholders, directly informing go-to-market strategy for community-solar models.
Selected work
Hermes / personal agent system: applied AI workflows with scoped agents, memory, tools, project context, Telegram, scheduled automations, and voice summaries.
Energy Scout: solar ROI and incentive decision support.
Real Estate Lead Engine: Verified listing → creative → Meta ads → capture → CRM → follow-up → signed lease, measured end to end.
Bitcoin ML: supervised ML and quant research across macro, price, and on-chain data.
Agentic workflows — Designed and operate a coordinated AI development/workflow system with Claude Code, Hermes-style agents, model routing, persistent context, scheduled jobs, Telegram, and voice summaries.
Used-car ML pricing model (MIT ADSP capstone) and Hull Tactical Kaggle competition: feature engineering, Random Forest/XGBoost, walk-forward validation of market "micro alpha."