I help build systems, teams, and software at the intersection of AI systems, product, and evaluation.
- Contract @ Handshake AI — evaluating ChatGPT app development and agent-built software across engineering, logic, design, and product quality
- Founder, Alloway LLC — applied ML products, analytics tooling, and autonomous agent systems
- B.S. Information Science, magna cum laude · M.S. Artificial Intelligence candidate — University of South Florida
I care about things that still hold up when someone looks closely & over time.
Browser automation is the real agent benchmark — a practical write-up and checklist for evaluating agents against the web that actually breaks them: auth, iframes, Shadow DOM, rich-text editors, passkeys, and final-state receipts.
Languages
AI / ML
Backend & APIs
Frontend
Tooling
🪙 SOLVENT — Self-Funding AI Agent
Built for the NVIDIA × Stripe × Nous Research Hackathon
An AI agent that runs as a profitable business. It sells research briefs, collects payment via Stripe, provisions its own compute using its own revenue, and refuses any job that doesn't clear a margin. A fully automated company with its own balance sheet.
Client pays Stripe → Agent earns → Agent fulfils → Agent pays vendors → P&L booked
| Metric | Value |
|---|---|
| Net profit (demo run) | $209.65 |
| Margin | 94% |
| Sponsor stack | NVIDIA Nemotron · Stripe Issuing · NemoClaw guardrails |
⚖️ juryrig
Zero-dependency Python toolkit for auditing LLM-as-judge pipelines — position-bias and verbosity-bias detection, judge panels with agreement scoring, and calibration (Brier, ECE) against human labels.
Full-stack sports analytics platform with live ML predictions, Kelly-based sizing, and a real deployed surface at aiadvantagesports.com.
📐 nba-ratings · kelly-js · matrix-rain
Flagship libraries behind the stack: Elo/logistic/Kelly primitives on PyPI as nba-edge, zero-dep TypeScript Kelly math on npm, and a retina-aware Matrix rain React component.
| Repo | One-liner |
|---|---|
| nba-ratings | Elo, win probability, calibration — published as nba-edge |
| kelly-js | Kelly sizing, CLV, bankroll stats — TypeScript, zero deps |
| matrix-rain | Digital rain component with reduced-motion support |
More projects and demos: ianalloway.xyz
- Autonomous agent systems with real economic loops (earn, spend, book P&L)
- Evaluation tooling that makes model and app behavior inspectable and trustworthy
- Applied ML products where the evaluation layer is part of the system, not an afterthought
- Sports decision systems — ratings → edge → Kelly → live product (AI Advantage)
other agents can spend money · SOLVENT runs at 94% margin
