- 🔭 I’m currently working on a Reinforcement Learning agent for 2048 and GAN variants for the “I’m Something of a Painter Myself” challenge.
- 🎓 M.S. CS (Machine Learning) @ UT Knoxville; B.S. CS @ Christian Brothers University (NCAA soccer alum).
- 🧠 Interests: ML systems, medical AI, NLP/RAG, RL, MLOps, HPC.
- 📍 Knoxville, TN (open to relocate) • 🇺🇸 Work authorization: U.S. citizen/authorized.
- 📫 Reach me: Cameronrader11@outlook.com
Lets connect: www.linkedin.com/in/cameron-rader-b69b2b1ba
- Medical Imaging: Built an end-to-end MRI knee abnormality classifier (MRNet) reaching ~0.91 AUC on validation with attention and transfer learning.
- Search & QA: Fine-tuned BERT for QA with custom RAG pipelines on long-context text (“A Study in Scarlet”).
- Full-Stack + Optimization: Shipped an EA FC 24 SBC solver (Flask + Next.js + Selenium + Genetic Algo) to cut solve time from minutes to seconds.
- HPC & Reliability: Contributed to UTK’s HPSC migration from Gold → ColdFront, integrating different technologies that come along with HPC management like Slurm, Ldap, etc.
- Content Automation: Built an AI video pipeline that hit 3.1M views and 12K+ followers in <30 days.
Repo: https://github.com/Cameronr11/MRNet-DeepLearning
What it is: Deep-learning pipeline to detect ACL/meniscus/abnormalities across axial/coronal/sagittal views.
How it works: Pretrained CNN backbones (ResNet/DenseNet), attention-weighted slice pooling, augmentation scheduling, ensemble across views.
Results: ~0.91 AUC on validation; class imbalance handled with weighted losses and stratified sampling.
Stack: PyTorch, torchvision, Optuna, TensorBoard.
Repo: https://github.com/Cameronr11/BERT-Novel-FineTune
What it is: Fine-tuning bert-base-uncased for extractive QA with custom RAG to overcome token limits.
How it works: Sentence/scene chunking, retrieval pipeline, and QA fine-tuning notebooks; install script for repro.
Stack: Hugging Face Transformers, Python.
Repo: https://github.com/Cameronr11/SBCAUTO
What it is: End-to-end system that scrapes inventory and uses a genetic algorithm to build valid, cost-efficient squads meeting SBC constraints.
Stack: Next.js (TypeScript, Tailwind, Chakra UI) + Flask API, Selenium, SQLite/Pandas, Socket.IO (real-time progress).
Impact: Reduces solve time from ~10 minutes to <45 seconds (typical runs).
Repo: https://github.com/Cameronr11/2048_RL
Focus: DQN/dueling/CNN variants; training curriculum aimed at consistent 1024→2048 tiles; symmetry augmentation & target update schedules.
Stack: PyTorch, Gymnasium-style envs, NumPy.
Status: Hardening training loop & evaluation harness.
Repo: https://github.com/Cameronr11/GAN-Variant-Research
Focus: Multiple GAN variants for style transfer & generation; local + cluster training workflows.
Status: Private while polishing baselines and evaluation; will open with reproducible configs.
What it is: Migration from legacy Gold to ColdFront for UTK’s HPSC portal
Impact: Smoother access management and more stable workflows for SLURM-based research jobs.
Role: Graduate Research Assistant (HPSC).
Stack: Python, SLURM, plugins, infra scripting (details redacted).
Graduate Research Assistant — High-Performance Scientific Computing, UTK
Aug 2024 – Present
- Supported researchers with job orchestration, SLURM troubleshooting, and workload optimization on multi-GPU clusters.
- Unified our project/access management with the HPC stack and standardized software installs (modules, dependencies, CUDA stacks), enabling faster onboarding, consistent runs, and fewer support escalations.
Software Developer Intern — Product Discovery, AutoZone
May 2023 – Aug 2023
- Contributed to backend modules powering semantic search used by millions of customers.
- Built an automated unit-testing framework to reduce deployment errors; improved synonym handling for better precision.
- Improving my 2048 RL evaluation and training throughput.
- Researching different variations of the GAN architecture
- Helping UTK researchers sharpen SLURM workflows.
- Side Project: Working with the new Wan2.2 Animate 14B model to upgrade my Tiktok Automation pipeline for better Ai related content