// projects / architecture index

Systems explained by
how they work.

The figures are snapshots from the case studies, rather than live counters. These are simplified maps of the boundaries, flows, and decisions inside 6 systems I have built, including archived experiments.

6 systems4 architecture layers eachCase studies attached.

Project / 01

KubeFit — Practical Kubernetes Training

Problem boundary

Knowing the command is not the same as executing under pressure. KubeFit grew out of the gap between understanding Kubernetes and reliably diagnosing, fixing, and verifying a real system.

Architecture / 01System view · simplified
01Read
◆Concepts guides
◆Playbooks
02Drill
◆Local cluster or VM
◆Timed scenario
03Grade
◆Real system state
◆Hints + feedback
04Repeat
◆Reports + notebooks
◆Targeted practice
Read at kubefit.com. Drill in your terminal. Grade real system state. Use the report to choose the next rep.
CLI training surfaceLocal executionState gradingFree publicly available
Inspect constraints + evidence →

Project / 02

TechBearing — Operator OS (Archived)

Problem boundary

Operational knowledge was scattered across infrastructure, repositories, telemetry, notes, and runbooks—and repeatedly lost between sessions. Operators and agents needed one durable, queryable view of what changed, what mattered, and what happened next.

Architecture / 02System view · simplified
01Observe
◆Infrastructure
◆Repositories
◆Telemetry
02Ingest
◆Source adapters
◆Event pipeline
03Remember
◆Operational state
◆Decision history
04Act
◆MCP over SSE
◆Operator + agents
Operational context becomes durable state, then inspectable tools for operators and agents.
~14 servicesMCP/SSE agent boundaryDurable state modelArchived status
Inspect constraints + evidence →

Project / 03

Bricktelligence — The AFOL Companion

Problem boundary

Serious collectors live across a stack of databases — Brickset, BrickLink, Rebrickable, Brickeconomy. They are catalogs: look something up, leave. None of them know your collection, your build habits, or whether the set you are eyeing is actually worth it for you. The hobby was drowning in reference tools and had no companion.

Architecture / 03System view · simplified
01Capture
◆Vision intake
◆Collection events
02Learn
◆Narrative facts
◆User aggregates
03Reason
◆Archivist tools
◆Verdict engine
04Experience
◆Web
◆iOS
◆Discover
Collection activity becomes personal memory that grounds every recommendation and surface.
iOS shipped product5 connected surfaces
Inspect constraints + evidence →

Project / 04

The Hive — Homelab Platform

Problem boundary

Cloud-native systems only become intuitive when you live with their failure modes. The goal was to build a durable environment where networking, storage, delivery, identity, observability, and application operations could be learned as one connected system.

Architecture / 04System view · simplified
01Enter
◆DNS + identity
◆Traefik + Kong
02Run
◆9-node k3s
◆Argo CD + Cilium
03Persist
◆Longhorn
◆TrueNAS / ZFS
04Compute
◆Proxmox
◆Alpha-5 / vLLM
Traffic, workloads, state, and inference stay separated while operating as one owned platform.
9 k3s nodes25+ GitOps applications102 TB ZFS storage32 GB GPU memory
Inspect constraints + evidence →

Project / 05

JobArchivist — Career Intelligence System

Problem boundary

A serious job search produces more state than a spreadsheet can hold: role versions, recruiter context, tailored documents, interview logistics, outcomes, follow-ups, and lessons that should change the next application. The real problem was not tracking rows. It was preserving context and turning it into better decisions.

Architecture / 05System view · simplified
01Capture
◆Jobs + feeds
◆Notes + interviews
02Normalize
◆Companies + contacts
◆Lifecycle events
03Remember
◆Documents + profile
◆RAG knowledge base
04Act
◆Pipeline + calendar
◆Grounded assistant
Messy job-search signals become normalized records, retrievable history, and actions that survive the browser tab where they began.
16 records in this case studyUsed during my search
Inspect constraints + evidence →

Project / 06

The Archivist Stack — Open-Source AI Substrate

Problem boundary

Three unrelated products kept converging on the same architectural shape. Rebuilding that substrate from muscle memory was useful evidence, but wasted motion. The reusable machinery needed to become inspectable, adoptable, and safe to build on.

Architecture / 06System view · simplified
01Model
◆Owned entities
◆Lifecycle state
02Record
◆Append-only events
◆Ingestion adapters
03Retrieve
◆Document chunks
◆Vector search
04Ground
◆Facts + tools
◆AI persona
A reusable substrate turns domain events into retrievable memory and grounded agent behavior.
Apache-2.0 license
Inspect constraints + evidence →