For the philosophy
Deterministic Decision Making
The core essay behind my current engineering philosophy: making decision systems reliable, inspectable, and reproducible in messy real-world environments.
Read essayWriting
Technical essays, project write-ups, and public learning logs.
I use writing to document how systems fail, how projects evolve, and what I learn while building. Some posts are polished essays, some are project series, and some are public notes from experiments.
A recurring theme is reliable decision-making under real-world uncertainty: prompts, tools, timing, data, edge devices, and systems that need to be observable after they fail.
Start Here
Three entry points depending on what you want to understand first.
For practical AI tooling
PromptLedger Series
A 5-part build series about turning prompt history into a local-first, inspectable workspace.
Open full seriesFor systems under constraints
Edge AI Sound Classifier
A 3-part build series showing how models behave when they leave notebooks and face real data, firmware, and constrained hardware.
Open full seriesFeatured Writing
Selected pieces that best represent the themes behind the work.
Deterministic Decision Making in Non-Deterministic Environments
The central essay connecting my projects around reliability, observability, timing, data control, and deterministic system design.
ReliabilityObservabilitySystem Design
Read essay
Building a Local PDF Summarizer with LLMs — In Under 24 Hours
A fast local LLM build showing how lightweight tools can turn documents into usable question-answering and summarization workflows.
Local AIDocument QARapid Build
Read build note
I Intentionally Built a Bad Decision System (So You Don’t Have To)
A small benchmark-style project showing how hidden state, side effects, and weak system design can break decision reliability even when code appears to work.
Hidden StateSide EffectsDebugging
Read systems note
DataLens: A Read-Only Image Dataset Sanity Checker
A read-only dataset inspection tool for detecting corrupted images, duplicates, label alignment issues, and silent data problems before training.
Data QualityInspectionComputer Vision
Read data note
Product Build Logs
Release notes and build stories from tools that became public products.
OpenAnima v1 — Open-Source Desktop Overlay Engine for Windows
A release and build note for OpenAnima as a Windows desktop overlay engine.
Open SourceOverlay Engine
Read release note
OpenAnima v0.2 Preview — Turning the Windows Desktop into a Living Canvas
A preview note showing the evolution from a simple overlay idea toward a more flexible desktop asset system.
Creative ToolAsset System
Read preview
Project Series
Multi-part build logs where the process matters as much as the result.
5-part series
PromptLedger Series
A local-first prompt version control series covering prompt history, labels, diffs, review workflows, faster logging, and the workspace dashboard.
Start with Part 1
01 Local-first prompt version control
02 Labels, Status, and Better Diffs
03 A practical review workflow
04 Faster logging and organization
05 Local workspace dashboard
View full series
3-part series
JumpNet Series
From raw gameplay data to a real-time vision-based game agent.
Start with Part 1
3-part series
Edge AI Sound Classification
Dataset collection, feature extraction, training, and firmware deployment for an embedded audio classifier.
Start with Part 1 View full series
3-part series
Edge AI Trend Alarm
A Raspberry Pi Pico trend-detection project covering data gathering, training, and firmware integration.
Start with Part 1 View full series
2-part series
Modular Snip Recorder
A data collection and dataset inspection tool for behavior cloning workflows.
Start with Part 1 View full seriesArchive Series
Longer practice series and older learning logs kept compact.
3-part series
Simple Shell in C
A beginner-friendly systems programming series about building a simple shell in C.
View full series
22-part series
Daily LeetCode Challenge
A long-running practice series documenting daily LeetCode solutions and problem-solving notes.
View full seriesReading Paths
Choose a theme and follow the related notes.
Path 01
AI Workflow Reliability & Tooling
Prompt workflows, local LLM utilities, retrieval experiments, and debugging practices for inspectable AI systems.
Path 02
Data Quality & Inspection
Tools and practices for finding data problems before they become model problems.
Path 03
Edge AI & Hardware Constraints
Models, state machines, and firmware built for constrained devices.
Path 04
Perception, Timing & Computer Vision
Visual systems where data collection and response timing shape behavior.
Path 05
Generative & Representation Learning
Experiments with latent spaces, reconstruction, and conditional generation.
Path 06
Classical ML & Experiments
Feature engineering, model comparison, regression, and educational experiments.
Archive
Older learning logs, practice notes, and smaller experiments.
Tackling LeetCode Bash Problems
Bash · Unix Tools · Practical Scripting
External Archive
More on dev.to
This website highlights selected writing and grouped project series. The full stream of posts, release notes, experiments, and practice logs lives on dev.to.