Things I built on my own time, what I am digging into right now, and what I wrote down along the way.
A desktop tool that works out exactly how many shares to buy or sell to bring a portfolio back to its target weights. Live prices come in through libyahoofinance; the interface is Qt. I wrote it because I wanted to rebalance my own holdings without a spreadsheet.
github.com/awe31402/StockRebalanceA lightweight C++ library for robot arm control: the Denavit–Hartenberg parameter model, forward kinematics, and inverse kinematics. Header-light and dependency-free, so it can drop straight into an embedded project rather than pulling in a full robotics stack.
gitlab.com/awe31402/modern_roboticsThese are study repositories, not products — notes, exercises, and runnable code from books and courses. I keep them public because how someone learns says as much as what they have shipped.
Working notes on Running Linux Kernel, volumes 1 and 2 — 340 questions across processor architecture, ARM64 in the kernel, memory management, process scheduling, concurrency, interrupt handling, and crash analysis. Includes a practical guide to the crash tool: Kdump setup, a command cheat sheet, and eight real debugging walkthroughs.
github.com/awe31402/linux-interviewWhere I am pointing my next few years: edge AI and model optimization. Course work from MIT's efficient machine learning material, model lightweighting notebooks, and close readings of the papers underneath — why memory bandwidth, not raw compute, is what actually limits a model on real hardware.
github.com/awe31402/deep_learningTen chapters of runnable code for shrinking large models and getting them served: quantization and pruning pipelines, ONNX conversion, TFLite on the edge, and FastAPI deployment. Each chapter is set up to actually run — there is a setup script and a verified environment, not just snippets.
github.com/awe31402/model-light-weight-codeThings I dug into and wrote down, mostly about what the kernel is actually doing underneath.