Notes on control and machines that learn.
A working notebook at the edge where classical control theory meets modern machine learning — written for engineers who want the math, the code, and the intuition in the same place.
Topics
01 / ExploreControl Systems
PID, state-space, frequency domain, stability — the math that keeps things in line.
→Motor Control
DC, BLDC, stepper. FOC, commutation, encoders, current loops.
→Robotics
Path planning, motion control, manipulation, and the geometry of getting somewhere.
→Machine Learning
From linear regression to gradient boosting — the classical toolkit.
→Neural Networks
Architectures, training dynamics, and the bits underneath the hype.
→Reinforcement Learning
Agents, rewards, policies. Where control theory and ML meet.
→Tools
Linux, git, terminals, dotfiles — the practical tooling that makes engineering work.
→Stock Market
Markets from first principles — what moves prices, reading a company, portfolios, psychology, crashes, and forecasting.
→Recent writing
02 / Latest-
Aug 30, 2026Control SystemsCompanion to Letters Before Numbers, in 20 sections with 12 interactive figures. The residual from an analytical redundancy relation detects faults cleanly in noise-free simulat...
-
Aug 30, 2026ToolsA field guide to computer algebra for engineers, in 47 sections with 37 interactive figures. What a computer algebra system actually stores, the price of numbers versus the pric...
-
Aug 29, 2026ToolsEvery method for cutting token consumption in LLM and agentic pipelines — prompt caching, semantic caching, context compaction, prompt compression, retrieval instead of stuffing...
-
Aug 28, 2026Control SystemsEvery root cause investigation is an argument: someone claims that if a particular thing had been different, the failure would not have happened. A working guide to making that ...
-
Aug 21, 2026ToolsDoes working with frontier AI demand a high IQ? The evidence says something subtler: AI levels routine work and amplifies frontier work, and past the ability threshold the bindi...
-
Aug 10, 2026Stock MarketA field guide to reading price charts. Every candle compresses a session of buying and selling into one shape: four prices, wicks as rejection marks, and location that matters m...
-
Jul 6, 2026Machine LearningWhy plain self-attention chokes on a million-point mesh (quadratic cost), and how Transolver's Physics-Attention fixes it: softly assign each mesh point to one of a small fixed ...
-
Jul 6, 2026Machine LearningHow DoMINO (NVIDIA's Decomposable Multi-scale Iterative Neural Operator) predicts aerodynamics directly from a car's STL/point cloud. It encodes the whole geometry into a global...