ML Engineer interested in agent systems, machine learning infrastructure, and building things from first principles.
Currently building agentic-rag-eval-harness, an evaluation and regression testing framework for LangGraph agents with trajectory scoring, RAGAS evaluation, LLM-as-a-judge, and CI/CD integration.
At Stryker, I work on the iOS / Smart Automation team building on-device AI with CoreML and VisionOS. I also contribute to an agentic multi-agent testing platform for functional, security, and Python code validation.
Previously built Procuity, a clinical ML pipeline for pressure injury prevention using sensor-based movement classification with an ensemble of MLP, LSTM, CNN1D, and CatBoost models.
Built Pixelflow, a node-based AI image editor exploring hybrid CLIP and color-statistics optimization. Although I paused the research direction, it remains one of my favorite projects.
Currently building nanotorch, a minimal deep learning framework implementing tensors, autograd, neural network primitives, FlashAttention, and CUDA kernels from scratch.
