Machine Intelligence · Systems Builder
Engineering intelligent systems at the intersection of deep learning, scalable infrastructure, and applied research. Building models that don't just predict — they understand.
Core Systems
Designing and training large-scale neural architectures — from transformers to diffusion models — with a focus on efficiency and production-readiness.
Building robust pipelines that move models from experiment to production with zero guesswork — CI/CD for ML, model serving, and experiment tracking.
Crafting high-throughput data systems that feed models with clean, structured, and semantically rich input at any scale.
Bridging the gap between academic research and real-world impact — reproducing papers, adapting methods, and shipping results that matter.
Architecting the machinery around intelligence — APIs, microservices, real-time inference, and distributed training clusters that scale without breaking.
Selected Work
A dynamic model serving system that auto-selects the optimal model architecture and precision level based on incoming request context and hardware availability. 3× throughput improvement over static baselines.
View on GitHubAn automated feature engineering library that uses meta-learning to discover high-signal feature transformations from raw tabular data. Cuts feature engineering time from days to minutes.
View on GitHubResearch implementation of a lightweight diffusion model pipeline optimized for edge deployment. Achieves 90% parameter reduction with less than 5% quality degradation on benchmark datasets.
View on GitHubReal-time model evaluation framework that monitors production ML systems for data drift, concept shift, and performance degradation — alerting before failures cascade.
View on GitHubThe Human System
I'm a Machine Learning Software Engineer who builds things that think. My work lives at the boundary between research and engineering — where ideas from papers become systems that run in the real world.
I believe the best ML systems are those that are obsessively understood by the people who build them. Not black boxes dragged into production, but intentional architectures where every design decision has a reason.
When I'm not training models or debugging pipelines, I'm reading papers, contributing to open source, and exploring the edges of what's computationally possible.
Transmission Open
Whether you're working on a challenging ML problem, want to collaborate on research, or just want to talk systems — I'm always open to a good signal.
View All Projects