Morgan Stuart

I build agents, and the evals that show whether a change made them better.

I'm an applied AI engineer from Brooklyn and a senior at Cornell. I build agent harnesses, MCP servers that connect agents to real data, and a memory graph that gives my own agents context on me. I work mostly in Python and TypeScript.

I've spent two summers at InstaLILY, a startup that is now Series B, most recently as an AI and RL harness engineer intern on Lily, the company's coding agent. For InstaLILY's public research report, I led the evaluation work in Part 1 and helped synthesize the final draft. Part 1 reports a 276-trial head-to-head between Lily and Claude Code. At Cornell I'm president of Cornell Generative AI, a student engineering club of 40+ members. What I build on my own lives under Stuart Labs.

One day I want to start a company. This fall I rebuilt the club's technical interview around agent design (the cycle brought in 13 new engineers) and led its move from client work to its own products. On my own, with coding agents, I built Outreach CRM over about twelve weeks, a multi-tenant networking CRM with 52 MCP tools. It is in a private pilot: I use it for my own job search, and the club, its first organization, runs its outreach from it.

See the workGet in touchGitHubLinkedIn
Morgan Stuart on a New York City rooftop

Now

After I graduate in May 2027, I want to join an early-stage startup or a small team in New York or San Francisco and work on agents, harnesses, evals or RL environments. I want a team small enough that I own real problems and hear from users quickly.

  • ego-OS, my personal context engine. It serves a memory graph of my notes and work to every agent I run over MCP, and v0.1.0 is open source.
  • Outreach CRM, a networking CRM with its own MCP server, in a private pilot with Cornell Generative AI as its first organization.
  • Leading Cornell Generative AI through its first semester building its own products, after client work for QuickFi and Investcorp.
  • HIVE, a two-person Cornell CS 4701 project on self-improving agent harnesses. An evaluation gate decides which proposed harness edits stay. The scaffold runs end to end in dry runs, and real experiments come next.

Selected work

All work →

Stuart LabsOpen sourceJune 2026 to present

ego-OS

A memory graph, stored on my own machine, that turns my notes, meetings, email and agent planning docs into memories, entities and concepts, and serves them to any AI agent over MCP. Every retrieval change is benchmarked, and one benchmark led me to delete a feature I had just built.

Read the write-up →

InstaLILYCase studySummer 2026

Lily-Eval: turning a codebase's merged PRs into a resettable environment for a coding agent

An evaluation environment for Lily, InstaLILY's coding agent, built by rewinding production repos to just before real merged PRs and grading the agent's blind reimplementation. It ran a 276-trial head-to-head against Claude Code and became the reward signal for tuning Lily's harness.

Stuart LabsProductAug–Oct 2026, ongoing

Outreach CRM

A multi-tenant CRM for student networking outreach. AI drafts each message from facts on file, the student sends it, and the whole pipeline doubles as an OAuth-protected MCP server that their own agent can work.

Cornell Generative AICase studySpring 2025 – present

Running Cornell Generative AI like a startup

As president of Cornell's student AI engineering club, I rebuilt how we hire, delivered client work for two finance firms, led the move to our own products, and built the club's operations platform with coding agents.

Stuart LabsOpen source2025–2026

HomeAssist and Tool-Calling Mini

An always-on voice assistant that I rebuilt from a Raspberry Pi prototype into an interruptible, proactive agent on the OpenAI Realtime API. Tool-Calling Mini is my QLoRA fine-tune of Qwen3-4B that can take over its tool routing from a self-hosted server.

Experience

More about me →

  1. Summers 2025 and 2026

    InstaLILY · Software Engineer Summer Associate (2025), then AI + RL Harness Engineer intern (2026)

    Wrote harness extensions for subagent and multi-agent delegation in Lily, the company's coding agent, alongside the small senior team that built it on LangChain DeepAgents. Led the design of Lily-Eval, the evaluation environment behind Part 1 of InstaLILY's public research report, which I built with a few colleagues; my main piece was the package that turns a repository's merged pull requests into graded tasks. In my first summer I was the sole developer moving a legacy sales assistant onto MCP.

  2. Spring 2025 – present

    Cornell Generative AI · President (earlier lead engineer and New Member Education chair)

    Joined as an engineer in spring 2025 and designed the new member curriculum before becoming president. I now run the 40+ member club, including recruiting, client and sponsor work, and its fall 2026 move to self-led products.

Writing

Stuart Labs directory

Contact

Building something early?

If you're building agents at an early-stage company, I'd like to hear about it. Email is the fastest way to reach me.

morgan@morgannstuart.com