About

Sean Traynor

AI Systems Engineer. Tufts CS, 2024. 2+ years designing and deploying production multi-agent systems, RAG pipelines, and LLM evaluation infrastructure across property tech, neuroscience, and education. Founder of UMANO and Sandbox Labs.

Download Resume (PDF) Get in touch

Background

I'm a New York-based AI Systems Engineer who designs and delivers production AI systems for enterprise stakeholders. With 2+ years of hands-on experience across property tech (MakenaAI), neuroscience research (Columbia University, published in The European Journal of Neuroscience), and AI evaluation (Scale AI), I specialize in translating complex business requirements into scalable multi-agent architectures.

My flagship project is the Master OS: a custom AI operating system I designed and operate across seven business domains, spanning 9 git repositories with Phase 6 hybrid orchestration, trust boundaries enforced by PreToolUse hooks, and a daily distillation pipeline that compounds institutional knowledge into a durable wiki. 336+ pages. A 19/19 system_doctor grade.

Before focusing full-time on AI infrastructure, I worked across the full stack at MakenaAI as a sole engineer on mobile, AI, and AR projects — shipping a subscription platform to 2,000+ clients in week one and leading the "Digital Clone" ARKit pilot for remote property inspections. I'm a Tufts University CS graduate (2024) and 2019 NCAA National Champion in soccer. That background matters: I know how to ship systems and how to win under pressure.

Operating philosophy

01
Correctness over velocity. A system that breaks under load is worse than no system at all. I build invariants into hooks, not prompts — because invariants need to be enforced by the runtime, not hoped for by the model.
02
Architecture follows runtime constraints. Theoretical elegance is cheap. The right design is the one the actual system supports. I build from discovered constraints, not whiteboard diagrams.
03
Compounding knowledge is the moat. Systems that get smarter every day compound faster than static systems. Distillation pipelines, verified knowledge bases, and durable audit trails are structural advantages.
04
One command, intelligent cadence. Good systems have minimal ceremony. If it takes three commands and a manual checklist to do a thing, it won't get done consistently. I optimize for the one-command path.

Technical depth

AI / Agentic Systems

  • Claude Code orchestration and hook design
  • Multi-agent coordination and authority delegation
  • Prompt engineering and advisory/deterministic separation
  • FSM-validated state machines for agent workflows
  • Verifier/executor patterns with pre-merge gates
  • Knowledge distillation pipelines

Infrastructure

  • Multi-repo git topology and worktree management
  • CI/CD pipelines (pytest, ruff, system health checks)
  • Vercel deployments and static site optimization
  • Python scripting and automation
  • JSON schema design and validation

Mobile / iOS

  • Swift / SwiftUI / Objective-C
  • ARKit / RealityKit (AR pilot, Digital Clone at MakenaAI)
  • StoreKit (subscription & payment systems)
  • XCTest (unit and UI testing)
  • REST API integration (Flask, FastAPI backends)
  • App Store submission and compliance

Product & Domain

  • Marketing automation (Meta Ads API integration)
  • EdTech product development (Sandbox Labs)
  • Music industry operations (UMANO)
  • Algorithmic trading infrastructure (isolated / read-only)
  • Technical writing and systems documentation

Front-End

  • Astro, React, Next.js
  • TypeScript / JavaScript
  • C++20 / JUCE (audio DSP, real-time systems)
  • CSS architecture and design systems
  • Performance-first static site development

What I'm looking for

I'm targeting senior AI roles at frontier-AI companies — AI Engineering, AI Operations, and Solutions Architecture positions where systems thinking and production credibility matter more than résumé pattern-matching.

Companies I'm most interested in include Anthropic, Weights & Biases, Scale AI, Harvey AI, and Cognition AI — places where the work is deep, the bar is high, and the domain is advancing fast enough that building your own operating system is considered a reasonable weekend project.

Open to remote, hybrid, or NYC-based roles. Available to start within 2–4 weeks.

sptraynor2001@gmail.com LinkedIn GitHub