Sean Even · Notre Dame, Indiana · Available for engagements

Paper
to Product

I am a robotics PhD researcher at the University of Notre Dame and the engineer behind 777 Enterprises — mechanical and electrical design, CAD, machine learning, and software that actually ships.

Five peer-reviewed papers on one side of the desk. Eleven shipped products on the other. The same person on the whiteboard and in the repository — no handoff, no account manager, no strategy deck that somebody else has to build.

Response
Within one business day

Start an engagement

24h response

Sent directly to Sean · No mailing list

Peer-reviewed

5

IEEE ICRA · IROS ×2 · ISER · arXiv

Shipped

11

Games, apps, hardware, ML systems

Disciplines

3

Mechanical · electrical · computer science

Engineers on your project

1

Architecture through production

(01)

The Practice

Seven ways to hire the practice. The unusual part is not any one of them — it is that the same engineer covers the mechanism, the board, the model, and the interface, so nothing is lost in a hand-off between four vendors.

01

Mechanical Design & Product Engineering

Hardware that has to exist physically · Mechanism design · Manufacture

A degree in mechanical engineering and a decade of building the thing afterward. Machine design, thermodynamics, fluid mechanics, materials and manufacturing processes — applied to mechanisms, enclosures, and assemblies that get made rather than rendered. Past work runs from a 130-part mountain board designed from scratch to automated-storage logistics for a $1B food distributor.

  • Mechanism design
  • DFM
  • Tolerance & fit
  • Materials selection
  • BOM & sourcing
  • Thermal & fluids

02

Electrical & Embedded Systems

Boards, firmware, power budgets · Devices that run unattended

Electrical engineering minor, PhD in electrical engineering, and firmware in the field. Microcontroller systems on ESP32 and Arduino, BLE and USB-serial protocols, LED driving at scale, power budgeting, and the regulatory reading that decides what you are allowed to build. Two LED-panel products in this portfolio exist mainly as a protocol argument settled correctly — where to put the intelligence, and what the link can actually carry.

  • ESP32 · Arduino
  • BLE & serial protocols
  • FastLED / WS2812B
  • Power budgeting
  • Sensors & actuation
  • Regulatory review

03

CAD, 3D & Architectural Modeling

Parts, assemblies, buildings · Anything that needs to be drawn before it is made

Extensive CAD practice across mechanical and architectural work — parametric part and assembly modeling, large assemblies, drawings and GD&T, print- and CNC-ready output, and architectural and interior 3D modeling for visualization. Five years in SolidWorks, two in Onshape, plus the graphical-communication and materials coursework underneath it.

  • SolidWorks
  • Onshape
  • Large assemblies
  • Technical drawings
  • Architectural 3D
  • 3D print & CNC prep

04

Machine Learning & Applied AI

Models that have to work on your data, on your hardware

Deep architectures built and trained from first principles rather than imported from a tutorial — variational autoencoders, convolutional decoders, diffusion models, neural style transfer, and learned controllers. Published work applies machine learning to robot burrowing; shipped work trains a VAE on an artist's own paintings in three minutes on a laptop and validates the JavaScript decoder numerically against PyTorch, because a plausible picture is not proof of a correct model.

  • PyTorch
  • VAEs & autoencoders
  • CNNs
  • Diffusion
  • Style transfer
  • Training & evaluation
  • On-device inference

05

LLM & Agent Systems

Language-model projects that have to leave the demo

Production LLM systems with the unglamorous parts attached: structured outputs and schema validation, evaluation harnesses, cost and token control, graceful offline fallback, and rules enforced in code rather than in a prompt. One shipped game overrules the model when it misbehaves; one research tool runs a fifteen-persona boardroom whose grades are scored against the closing tape and fed to a small neural net that learns which personas to trust.

  • Structured outputs
  • Multi-agent systems
  • Retrieval
  • Evals & regression suites
  • Cost control
  • Offline fallback

06

Robotics, Control & Simulation

Locomotion, soft actuation, real-time control · The published specialty

The doctoral work: limbless and soft robotic locomotion, pneumatic actuation, real-time feedback control derived from partial differential equations, geometric mechanics and variational integrators, and machine-learning-driven burrowing in granular media. Published at IEEE ICRA, IROS, and ISER. If a mechanism has to move through something difficult — sand, soil, a pipe, a body — this is the exact problem.

  • Soft robotics
  • Locomotion
  • PDE feedback control
  • Geometric mechanics
  • Granular media
  • Sim-to-real

07

Software, Web & Mobile

The layer users actually touch · Shipped on five platforms

Full-stack delivery across an unusually wide platform range, all of it represented by something finished: Cloudflare Workers and Durable Objects, React Native, Flutter, SwiftUI, Unity, and plain dependency-free web apps that run by opening a file. Design included — every site and interface in this portfolio was designed as well as built.

  • TypeScript · Python
  • Cloudflare Workers
  • React Native · Expo
  • Flutter · Firebase
  • SwiftUI
  • Unity
  • Interface design

How an engagement runs

(02)

Selected Work

Four of eleven. Each one finished enough to hand to a stranger — which is the only standard that separates a portfolio from a folder of experiments.

(01)

Bablytics

AI market research · Python · 15-persona boardroom
bablytics.com

An AI boardroom that reads the market's infinite library twice a day.

Every weekday morning it scrapes options-volume spikes, US and world news, congressional trade disclosures, chart technicals and cycle data, builds roughly forty candidate trades, and convenes fifteen AI personas to grade each one. A Skeptic rebuts with every grade in hand; an Executive publishes a ranked Top 20 and a morning memo. After the close, the Evening Court scores the day's calls against the tape and each member journals what they got wrong — and those lessons feed the next morning's grading.

Realized 1-day and 5-day returns also train a small neural net that predicts outcomes from board grades, so the Executive learns which members to trust. Named for Borges's Library of Babel: the market publishes an infinite library every day, and nearly all of it is noise.

  • Python
  • Multi-agent LLM
  • Neural net
  • Data pipelines
  • Research tool

(02)

Blunderbuss

Chess · Cloudflare Workers + Durable Objects
Live

Chess with a friend over text, on a per-move clock.

Correspondence chess has two failure modes: the opponent who takes three days to reply, and the opponent who takes three days because they are feeding the position to an engine. A short per-move clock fixes both. Miss it and the server plays a random legal move for you — fired from a Durable Object alarm, so it happens whether or not anybody has the page open.

Each player's clock starts when they open the game, so nobody is punished for delivery. The random move is deterministic — seeded from the position and ply and nothing else — which is what lets any client redraw the whole board from the move list alone. No accounts; a player token in local storage is the entire identity model.

  • JavaScript
  • Durable Objects
  • Edge compute
  • Custom chess engine

(03)

Atelier

Local-first generative art · PyTorch · Desktop app

An apprentice for your studio — that never sends the artist's work anywhere.

Give it a few of an artist's pieces and describe what you want. A five-stage pipeline draws the content with SD-Turbo, repaints it with depth-weighted VGG19 style transfer at native brush scale rather than a squashed thumbnail, restores your composition, and — optionally — consults a variational autoencoder trained locally on that artist's own marks to ask whether the paint is in their vocabulary.

Private by construction. No account, no upload, no server; the only network access is a one-time model download. Nothing trains a shared model on anyone's work. Ships as a Mac, Windows and Linux desktop app.

  • PyTorch
  • VAE
  • Diffusion
  • Style transfer
  • Offline

(04)

Board Foxes

Unity 6 · 3D game · 8 attractions, 34 talking NPCs

Seven skateboarding foxes. One coastal resort island. Every attraction is a video game.

A hub world containing eight complete games — a score-attack sky park, a four-course kart grand prix, a percent-knockback platform fighter, a creature-catching sky archipelago, a 2D side-scroller, a 24-tile board game, a three-trial marble climb, and a full creature RPG with a 24-species dex, a ten-type chart, four gyms, an Elite Four and a post-game.

The resort runs a living world underneath: a twelve-minute day/night cycle, weather, staff work shifts, scheduled events, a fishing minigame, and a house you furnish. All thirty-four staff hold real conversations through an LLM, and fall back to authored persona lines with zero network when no key is present.

  • Unity 6
  • C#
  • Procedural scene generation
  • Game design
  • LLM NPCs

All eleven projects

(03)

Research

PhD candidate in electrical engineering at the University of Notre Dame, in the MiniRo Lab (Minimalist Naturally-Inspired Robotics), advised by Prof. Yasemin Ozkan-Aydin. The work is locomotion in hard places — soft bodies, granular media, and the mathematics that makes either tractable.

2025

Evaluating Tools for Collective Soil Excavation: Challenges and Insights

Sean Even, Omer Kurkutlu, Yasemin Ozkan-Aydin

ISER 2025 · International Symposium on Experimental Robotics

2024

A variational approach to geometric mechanics for undulating robotic locomotion

Sean Even, Patrick S. Martinez, Cora Keogh, Oliver Gross, Yasemin Ozkan-Aydin, Peter Schröder

arXiv:2409.10827 [cs.RO]

2024

Machine Learning-Driven Burrowing with a Snake-Like Robot

Sean Even, Holden Gordon, Hoeseok Yang, Yasemin Ozkan-Aydin

IEEE ICRA 2024 · International Conference on Robotics and Automation

Full publication list & academic background

(04)

Stack

Tools with something finished behind them. Nothing listed here is theoretical.

Mechanical & CAD

  • SolidWorks — parts, assemblies, drawings
  • Onshape — cloud parametric CAD
  • Architectural & interior 3D
  • GD&T and tolerance stacks
  • 3D printing & CNC preparation
  • FEA & hand calculation

Electrical & Embedded

  • ESP32 · ESP-IDF · Arduino
  • BLE peripherals & custom protocols
  • FastLED / WS2812B at panel scale
  • USB serial & framing
  • Power budgeting & automotive supply
  • PlatformIO

ML & Research

  • PyTorch — VAEs, CNNs, decoders
  • MATLAB — numerical methods, analysis
  • NumPy · pandas · scikit-learn
  • Diffusion & neural style transfer
  • PDE-based feedback control
  • Geometric mechanics & variational integrators

Software & Platform

  • TypeScript · JavaScript · Python
  • Cloudflare Workers · Durable Objects
  • React Native / Expo · Flutter · SwiftUI
  • Unity 6 · C#
  • Firebase · Firestore · Cloud Functions
  • C / C++ · Dart · Swift

Tell me what
you’re building

Mechanism, board, model, or interface — or all four. Send the problem and where it stands today. You will hear back from the person who would do the work.

Start an engagement Email directly