Backend, Systems & DevOps Engineer building Egg ("Don't crack") — a privacy-first software initiative designed to support executive function, task decomposition, and empirical work-pattern insight for individuals with ADHD.
My public repositories represent standalone engineering iterations where I developed and tested core architectural concepts—such as bounded-context local LLM decomposition and structured entity modeling—followed by Microservice-egg, where the lessons learned from those standalone projects are consolidated into a fully developed, production-grade microservice platform.
| Repository | Role & Engineering Focus | Stack |
|---|---|---|
Microservice-egg |
Unified Production Platform (Monorepo): Consolidates the lessons learned from earlier standalone projects into a full multi-service architecture. Currently implementing the Cognitive Telemetry & Activity Block Engine: a local-first 30-second micro-epoch ingestion daemon (Protocol-decoupled hardware hooks + --simulate mode for headless CI), lossless algebraic SQLite WAL storage ( |
Python 3.12 • FastAPI • SQLite WAL • uv • Docker • mypy --strict • Ruff • Pytest
|
egg-ai |
Standalone Iteration — Recursive LLM Task Decomposition: A standalone application solving small-context-window limits on consumer local LLMs. Identifies a core topic from a user query, breaks it into sub-topics, and issues independent bounded-context queries to build an ordered tree structure where leaf nodes become actionable execution steps. |
Java 25 • Spring Boot 4 • Docker • JSON Schema • Local LLM / OpenAI
|
workflow_assistant |
Standalone Iteration — 5W1H Entity & Workflow Modeling: A standalone application that models tasks, recurring workflows, and problems using 6 structured parameters (Who, What, When, Where, Why, How) and coordinates asynchronous AI check-ins. |
Java • Maven • GitHub Actions CI
|
ai-gatherer |
Standalone Iteration — Structured Concept Extraction: The initial prototype preceding egg-ai, built to validate strict JSON-schema LLM responses, HTTP retry resilience, containerized deployment, and security analysis. |
Java 25 • Spring Boot 4 • Docker • DevSecOps
|
-
Context-Scoped Monorepo & Microservices: Clear domain boundaries with isolated environments, reproducible lockfiles (
uv.lock/pom.xml), and containerization (Dockerfile). -
Automated Verification Gates: Strict static typing (
mypy --strict), fast linting and formatting (ruff), and$\ge 90%$ branch-coverage unit and integration tests (pytest) designed to run deterministically in headless CI runners. - Privacy by Construction & Bounded Contexts: Designing local-first telemetry with zero raw keystroke capture and structuring LLM pipelines to keep context small, predictable, and runnable on local consumer hardware.
-
Engineering Sprint Reports: Follow the sprint-by-sprint Architecture Decision Records (ADRs) and progress in
Microservice-egg/reports.