Ethos
Learn the stack until the abstraction feels physical.
Learning Atlas / Working notes
Making complex systems legible.
A holding room for fields I am trying to understand deeply — not as trivia, but as operating systems of power, capital, physics, software, and taste.
ModeVisual first
PaceSlow maps
OutputField guides
Concept rooms
Each card opens a domain — a distilled ethos, a mental model, and a visual companion designed to make the terrain easier to revisit. Two rooms are open; more are queued.
01 / SemiconductorsPhysics → Capital → Compute
The invisible terrain.
A semiconductor is not merely a chip. It is compressed geopolitics, precision manufacturing, memory bandwidth, energy, and software demand — etched into matter at the edge of what humans can reliably control.
Open graphic page02 / Agentic EngineeringAtomize · Integrate · Refine
The orchestrating loop.
AIR is not a builder. It is the seam-watcher above the specialists — atomizing a fuzzy ambition, dispatching agents in parallel, and re-planning the moment work returns. The discipline is a loop, not a pipeline.
Open working diagramNext rooms
- Private CreditDrafting
- AI InfrastructureQueued
- Luxury SystemsNoted
- Urban HospitalityNoted
Outputs
Where the working notes turn into deliverables. Each output is a self-contained case study — the work itself plus the way it was made.
01 / Investment ResearchChip → Board → Rack → Grid
A nine-name cohort, built in a weekend.
Nine tickers across the chip-to-grid stack — semiconductors plus AI-power infrastructure — researched, refined, and published with one PM at the seam and a fleet of specialist agents underneath. The case study covers the work, the team that wasn't a team, and the productivity ledger.
Open case study02 / Investment ResearchUSDC · GENIUS · One Polarizing Ticker
One name, seven lenses, five tensions.
A single-name committee on Circle Internet Group (CRCL) — seven specialist analyst lenses dispatched in parallel, one synthesizer at the top, one PM at the seam. The case study covers the work, the disagreement-as-signal pattern, and the staged sizing playbook tied to dated catalysts.
Open case studyEthos
The leverage is in the loop, not in the tokens.
Outputs queued
- L4 memory deep-diveDrafting
- Coinbase 8-K parser · Aug 2026Queued
- Stablecoin competitive map · quarterlyDrafting
- L1 lithography deep-diveQueued
- Hyperscaler capex trackerNoted
- Tokenized-deposit threat trackerNoted
How I’m learning
The goal is not to collect links. It is to build enough internal structure that new information has somewhere useful to land.
Map the stack
Start with the layers, dependencies, bottlenecks, and feedback loops. A domain becomes clearer when it has a visible skeleton.
Name the tensions
Every interesting system has constraints: cost, power, time, regulation, taste, trust, or physics. The constraints reveal the strategy.
Return visually
Convert notes into diagrams and cards so the knowledge is not just stored, but navigable — something you can re-enter quickly.