Jared Behunin
AI isn't failing you technically.
It's failing you operationally.
I help mid-sized companies close that gap — not with more tools, but with the operational architecture to make AI actually run.
11 years building and operating technical systems at Amazon. Now doing it independently — the projects below are what that looks like in practice.
Work
Production AI Systems
Self Reinforced AI Trading Engine
A fully autonomous multi-agent paper trading system that scans the market nightly, generates directional signals, and executes paper trades without human intervention. Pythia collects and processes market data. Eris makes execution decisions, sizes positions, and manages risk. The system posts live trade updates twice daily to a public Discord server.
Self Managed AI Tuning Platform
An autonomous fine-tuning platform that continuously improves the directional signal model powering the trading engine. Chiron manages the full training lifecycle — hypothesis tracking, dataset construction, training runs, outcome debriefs, and next-cycle planning — without human orchestration. Tiresias is the fine-tuned model artifact it produces.
AI Invented Tax Deduction Finder
An AI system that applies deep knowledge of federal tax law to identify non-obvious deductions — ones a taxpayer wouldn't think to claim and an accountant might not surface unprompted. The system reasons across tax code, financial context, and individual circumstances to generate deduction hypotheses with supporting rationale.
AI Enhanced Reentry & Benefits Program
A platform that helps people navigating reentry from incarceration identify and access the benefits, programs, and resources they're entitled to. Built with a business-like, non-patronizing design philosophy — direct, competent, and respectful of the user's intelligence. All user data stays local; no external LLM calls with personally identifiable information.
Background
About
Head of Tech Operations at Amazon Business, a $40B business unit. Built and owned technical operations infrastructure, agentic automation workflows, and MCP architecture serving 2,000+ concurrent roadmap projects annually.
Previously Chief of Staff to the VP of Amazon Business, and before that spent eight years in senior program management roles across Amazon — including conceiving and launching Amazon Extra Large (AMXL) and leading global direct fulfillment and transportation programs.
Outside of Amazon, building production AI systems: autonomous trading engines, self-improving ML pipelines, and AI-native applications that solve real problems.
Stack