🔭 I specialize in building robust machine learning and software utilities and services.
🌱 I create high-quality libraries and tools that prioritize reliability, performance, and usability.
• PyTorch Algorithm Development: Optimizing CUDA operations for data processing and model operations (training/prediction) locally and in cloud environments
• MLOps Orchestration: Managing ML operations with and without CUDA in cloud for training, evaluation, and deployment
• Cloud ML Architecture: Designing reliable, highly available ML services with agility for change and updates
• Research Automation: Automating research and experimentation of machine learning in cloud environments
• LLM & Agent Technology: Deploying LLMs to GCP and Kubernetes, creating backend engineering servers for LLM streaming and MCP tool servers
Here's a collection of my open-source projects:
A Python package providing a collection of activation functions not implemented in PyTorch.
Deep learning architecture specifically for tabular data, combining interpretability and high predictive performance.
Python package designed to simplify analysis of tabular data for inference models with powerful diagnostics and interpretability.
Create probability distributions from Gymnasium spaces with a simple and intuitive interface for reinforcement learning environments.
Simplifies the conversion of pandas backends to pyarrow, allowing a seamless switch to pyarrow pandas backend.
Define Kubeflow Pipeline (KFP) Components with Python Dataclasses for dataclass-driven component definition.
A Model Context Protocol (MCP) server for agent development tools, providing scoped authorized operations in projects.