Data engineer forging AI agents, MLOps pipelines, and Rust tooling for ML. Author of From Pythonista to Rustacean.https://xavierforge.dev/https://xavierforge.dev/en/posts/github-pages-custom-domain/https://xavierforge.dev/en/posts/github-pages-custom-domain/The full process of putting a custom domain on an Astro + GitHub Pages personal site — choosing a domain, Cloudflare DNS, CNAME, the HTTPS certificate, and account-level domain verification.Thu, 28 May 2026 00:00:00 GMThttps://xavierforge.dev/en/posts/ironman-2021/https://xavierforge.dev/en/posts/ironman-2021/An honorable-mention entry from the 2021 iThome Ironman contest. Starting from the lifecycle of a machine learning product, it walks through the deployment, modeling, data, and scoping stages an ML product goes through from concept to launch, and builds a simple web app along the way.Fri, 15 Oct 2021 00:00:00 GMThttps://xavierforge.dev/en/posts/ironman-2023/https://xavierforge.dev/en/posts/ironman-2023/My completion recap and 30-day index for the 2023 iThome Ironman contest — exploring, from a Python user's perspective, whether Rust has a place in MLOps. The series later grew into the book "From Pythonista to Rustacean."Sun, 15 Oct 2023 00:00:00 GMThttps://xavierforge.dev/en/posts/jetson-nano-openclaw/https://xavierforge.dev/en/posts/jetson-nano-openclaw/Turn a retired Jetson Nano (Ubuntu 18.04) into a securely isolated local AI Agent sandbox: extend its life by enabling ESM, work around the glibc limit with Bun, and step through deploying the OpenClaw AI Gateway with a systemd service.Sun, 22 Feb 2026 00:00:00 GMThttps://xavierforge.dev/en/posts/numba-cuda-puzzles-1/https://xavierforge.dev/en/posts/numba-cuda-puzzles-1/Let's play a game and learn how to write CUDA kernels, the algorithms that power 99% of deep learning! Through Sasha Rush's GPU Puzzles and Python's Numba library, painlessly build an intuition for CUDA's thread hierarchy, kernel writing, and memory model.Mon, 17 Jun 2024 00:00:00 GMThttps://xavierforge.dev/en/posts/numba-cuda-puzzles-2/https://xavierforge.dev/en/posts/numba-cuda-puzzles-2/Use GPU Puzzles 9–14 to implement pooling, dot product, 1D convolution, prefix sum (Blelloch), axis sum, and matrix multiplication, diving into shared-memory optimization and parallel reduction to translate Python algorithms into efficient CUDA kernels.Sun, 23 Jun 2024 00:00:00 GMThttps://xavierforge.dev/en/posts/zerostack-memory-design/https://xavierforge.dev/en/posts/zerostack-memory-design/A design walkthrough for the memory subsystem I contributed to zerostack, a minimalist Rust coding agent. Surveys what Claude Code, oh-my-pi, memsearch, and opencode-agent-memory do, explains why vectors were declined, and lands on plain-Markdown memory with a search that catches up to vector retrieval for this use case.Sat, 06 Jun 2026 00:00:00 GMT