I'm a high school student learning machine learning from first principles — linear algebra, optimization, backpropagation — and building my way up to transformers by hand. I program robots for FRC Team 1086 (Blue Cheese), run interpretability experiments across open LLM families, and build full-stack systems designed to put models to work. Long-term, I want to contribute to AI research at the highest level, through work that is both theoretically sound and deployable.

$ cat about.sh
ROLE="Machine Learning Engineer in Training"
EXP="Student builder — on GitHub since 2024"
DOMAIN="AI/ML · FRC Robotics · Full-Stack Systems · Security"
STACK="Python · PyTorch · C++ · ROS 2 · TypeScript · Java"
OPEN_TO="research collaboration · internships · hard problems"

Languages

ML · Vision · Data

Web & Backend

Systems & Tools

2024 → now · FIRST Robotics Competition — Team 1086 "Blue Cheese" · Programming & Controls, Drive Team

• Built the team's CUDA vision stack on Jetson Orin Nano Super: Isaac ROS AprilTag detection → multi-tag SQPnP pose → NetworkTables 4 to the roboRIO
• Developed BlueScout, the team's offline-first scouting and strategy platform for the 2026 REBUILT season
• Wrote a real-time A* path planner with dynamic obstacle avoidance for driver-assisted navigation
• Generated synthetic 2D/6D pose training data in NVIDIA Isaac Sim for vision experiments
• Drive team member at competition — strategy and execution under match pressure

C++ ROS 2 CUDA Java WPILib Next.js PostgreSQL

2025 → now · Independent ML Research & Study

• Designing and running a probe-reliability benchmark across 6 open LLM families with geometric and spectral diagnostics, on local GPU and Lambda Cloud
• Working through the NeetCode ML course — implementing every stage from gradient descent toward a working GPT, assembled into neetcode-gpt
• Ongoing self-study of machine learning theory and research papers, with systems-level experimentation and benchmarking
• Studying transformer internals, efficient fine-tuning (LoRA, quantization), and ML systems design

Python PyTorch HuggingFace Linear Algebra Optimization

2023 · Center for Information Technology (CIT) — Deep Run High School

• Built "TaskMaster," a full-stack task-management web application as a CIT Windows Server project
• Clean-architecture ASP.NET Core 8 Web API — Core/Infrastructure/API layers, EF Core 9, AutoMapper, Swagger
• React 19 + Vite + Tailwind 4 client with dashboard, reporting, and export features over a task domain model with priorities, statuses, and due dates
• Deployed and served from the CIT Windows Server on IIS with SQL Server

C# ASP.NET Core EF Core SQL Server IIS React

2023 → now · CyberPatriot — National Youth Cyber Defense Competition

• Three-year competitor in the National Youth Cyber Defense Competition
• Reached Semifinalist standing
• Focused on Windows Server security hardening under competition time pressure
• System defense on scored competition images as part of a team

Windows Server System Hardening Cyber Defense

Coursework: AP Calculus BC · Multivariable Calculus · Discrete Mathematics · Math Structures · AP Physics · AP Computer Science A

learning:
  - transformer internals & efficient fine-tuning (LoRA, quantization)
  - ML systems design
  - probe-based interpretability methods

building:
  - probeReliabilityScript: probe benchmark across 6 open LLM families
  - neetcode-gpt: the transformer stack of a from-scratch GPT
  - bluescout: FRC 1086 scouting platform for the 2026 REBUILT season
  - rosvision: CUDA AprilTag vision on Jetson Orin Nano Super

exploring:
  - LLM interpretability research
  - synthetic data generation in Isaac Sim
  - research paper reproduction

open_to:
  - research collaboration
  - internships & mentorship
  - hard problems worth solving

Build. Understand. Improve.