Loading interactive logo…
Since March 2024, I have worked in AI training, moving from difficult mathematics and physics problems into quality control and robotics simulation. My independent work also explores structured visual reasoning, practical software, and music.
See the workI build simulation tasks, reference policies, diagnostics, and scoring for robotics training data. The work shown here includes autonomous aerial control, cooperative robotics, and precision navigation.
Explore robotics work
Loading interactive logo…
GraphDat is my AI-backed geometry system: a natural-language layer over GeoGebra that pairs a fine-tuned FLAN-T5 Base model with deterministic scene execution, keeping each construction inspectable, persistent, and editable.
Explore GraphDatI studied Music Technology at Montana State University and have been building home studios and recording my own music since I was fifteen. Most of my solo work starts at the drum kit before growing through bass, guitar, vocals, editing, mixing, and mastering.
Explore music and recordingsAn independent software project that makes vendor information easier to review and use. I am keeping its workflow, business context, implementation, and future direction private while it develops.
As AI systems improved, my work moved from difficult written reasoning tasks and quality control into simulated physical problems where a model must write code that turns observations into actions.
Scale AI
Created challenging algebra, calculus, and physics data for leading AI systems, then reviewed expert-authored work for mathematical correctness, justified reasoning, instruction fidelity, and client readiness. Review work also included graduate-level data-pipeline tasks.
Alignerr
Create and audit high-value training data across changing programs and domains, adapting quality-control methods as the models and tasks evolve. Current work extends that progression into MuJoCo simulations that ask AI systems to write effective robotics policies.
Independent development
Building simulation-based robotics training problems and reference controllers while developing GraphDat and Materials Cost Tracker as longer-term product systems.
When a recurring workflow problem begins costing time or attention, I usually turn it into a small tool or operating protocol.
External monitoring waits for decisive simulation or CI evidence, then wakes the exact task instead of spending model effort on repeated polling.
A background Python and SSH bridge shares text and image clipboard content between my two Macs while keeping credentials and live task state machine-local.
Staged protocols carry an idea through feasibility, implementation, physical proof, evaluation, review, and publication without losing the evidence or decision boundary for each stage.
Background notifications deliver progress, failures, and reviewable artifacts to my phone while long-running work continues away from the workstation.
I am an applied mathematics graduate working in AI training, robotics simulation, and independent software development.
Since March 2024, my work has progressed from difficult mathematics and physics training data through expert review and quality control into robotics tasks where AI systems write policies for simulated physical systems. My independent work includes GraphDat and a private vendor-intelligence project.
View professional backgroundMontana State University · 2023
Montana State University
Some project details and datasets are withheld to protect client work and unpublished R&D.
I am open to roles and collaborations involving AI training data, embodied AI, simulation, interactive systems, technical product engineering, audio technology, and games-adjacent tools.
