Context

A detailed description of how Claude Code has been set up and used as a personal second brain for a Director of Data Engineering at Meta. Written March 10, 2026 — after ~2.5 weeks of active use.


1. Structure — How It's Organized

The Foundation: PARA on Google Drive

The system uses the PARA method (Projects, Areas, Resources, Archives) as its organizational backbone. Everything lives in a single directory on Google Drive (~/gdrive on both machines), which means it persists automatically and is accessible from multiple devices and Claude instances simultaneously.

~/gdrive/
├── CLAUDE.md                         # 450+ line master instruction file
├── devserver_workflow_guide.html      # Visual reference guide
│
├── 00_inbox/                         # Capture and daily operations
│   ├── action_items.md               # Running task list (~95 lines, 7 categories)
│   ├── meeting_backlog.md            # Queued topics for recurring meetings
│   ├── incubator/                    # Ideas not yet started
│   └── notes/                        # Daily artifacts
│       ├── YYYY_MM_DD_daily_focus.md     # 10 generated so far
│       ├── YYYY_MM_DD_daily_reflection.md # 4 generated so far
│       ├── YYYY_MM_DD_week_ahead.md      # 3 generated so far
│       └── .kestrel_cache/           # Intelligence pipeline outputs
│
├── 01_projects/                      # 8 active projects
│   ├── 01_h1_26_roadmap_gsd/
│   ├── 02_dashboard_excellence/      # DashEx — most active
│   ├── 03_fifu_data_foundation/
│   ├── 04_eoy_performance_cycle/
│   ├── 05_data_strategy/
│   ├── 06_ai_analytics_transformation/ # Heaviest — 10 meeting files, 2 outputs
│   ├── 07_analytics_blog/
│   └── 08_ai_training_ramp/          # LoopHole + 30-day plan
│
├── 02_areas/                         # Ongoing responsibilities
│   ├── hpms/                         # Weekly highlights/priorities
│   ├── session_context/              # Cross-instance continuity (3 files)
│   └── verticals_ds_pillar_leads/    # Recurring meeting area
│
├── 03_resources/                     # Reference material
│   ├── people/                       # 32 colleague files
│   ├── tommy_leadership_guide.md     # Values, style, voice
│   └── workplace_groups.md           # Group IDs for posting
│
└── 04_archives/                      # Completed projects
    ├── 01_analytics_summit_2025/
    └── 02_messenger_de/

Total: ~100 markdown files, 1 HTML guide, across 8 active projects, 3 areas, and 32 people profiles.

How Projects Relate to Each Other

The projects aren't independent silos — they form an interconnected web:

Project Core Focus Key Connections
DashEx (#02) Dashboard quality standards Feeds into Data Strategy (#05); Nest-vs-Unidash impacts AI Transformation (#06)
AI Analytics Transformation (#06) AI-first analytics vision Parent theme for AI Training (#08); Configurable metric skills connect to DashEx
AI Training & Ramp (#08) Tommy's personal AI transformation LoopHole system is the prototype for Jeremy; 30-day plans feed back to org (#06)
EOY Performance Cycle (#04) Career conversations Career conversations with directs connect to AI expectations from #06 and #08
H1 Roadmap (#01) Team planning Roadmap now must reflect AI strategy from #06
Data Strategy (#05) Cross-pillar DE strategy Frames everything — DashEx, AI transformation, FIFU all ladder up here

The most active axis right now is #06 (AI Transformation) ↔ #08 (AI Training) ↔ #04 (Performance) — the AI push is simultaneously a strategic initiative, a personal development plan, and a performance expectation from Tommy's manager.

Each Project's Internal Structure

Every project follows the same pattern:

This keeps the CLAUDE.md scannable while preserving full detail in subfolders. Claude reads the CLAUDE.md first and digs into subfolders only when relevant.