Projects
- Boosters — A high-performance gradient boosting library for Python and Rust
MLOps engineer passionate about machine learning, energy forecasting, and building high-performance systems. I write about the things I learn.
Running the same ONNX models from Part 2 on Apple Silicon. From CoreML being 41% slower than CPU to hitting 28 FPS real-time through format flags,…
Taking ONNX models from 80 FPS (PyTorch CUDA) to 97 FPS (TensorRT mixed-precision), including a deep dive into a known TensorRT fusion bug that…
A foundation video segmentation model runs at 1.3 FPS. This post covers the first step toward real-time: getting it into ONNX, which sounds simple…
The story of building a gradient boosting library in Rust—from dissecting XGBoost models to achieving performance parity and beyond.
A practical guide to configuring gradient boosting: learning rate, tree depth, regularization, and systematic tuning strategies.
How decision trees fit into gradient boosting: from second-order Taylor expansion to the split gain formula that powers XGBoost and LightGBM.
When to choose XGBoost vs LightGBM: speed, accuracy, features, and practical recommendations based on your use case.
How LightGBM bundles sparse features to reduce memory 25×, and how both libraries handle categorical data without one-hot encoding.
How LightGBM speeds up training by keeping high-gradient samples and subsampling the rest. Plus row and column sampling strategies.
XGBoost grows trees level-by-level, LightGBM splits the best leaf. Learn when each strategy wins and how num_leaves relates to max_depth.
The optimization that makes gradient boosting fast: how histogram-based training reduces split finding from O(n log n) to O(bins) with the…
How gradient boosting optimizes in function space: from pseudo-residuals to the complete boosting algorithm, with derivations for common loss…
An intuitive introduction to gradient boosting: learn how combining weak models creates powerful predictions, from ensemble intuition to the boosting…
The blog introduces SageMaker as a versatile AWS service for tasks like building data pipelines and deploying machine learning models, addressing…
This series of blog posts aims to demystify the associated terminology and concepts, providing a comprehensive guide for individuals looking to…