Project case study

Draftboard OS

A digital, in-person snake draft board for commissioner-run fantasy football leagues.

Problem

Fantasy football draft nights still lean on spreadsheets or a shared screen someone has to babysit. Draftboard OS is a spec-driven MVP for a single, clear use case: an in-person snake draft that a commissioner runs from one board. It deliberately excludes auctions, pick-trading, remote pick submission, and third-party platform integrations — those are out of scope for this pass.

Approach

The backend is Python and FastAPI, with SQLAlchemy and Alembic for data and migrations, tested with pytest. The frontend is Next.js App Router. Postgres and Redis run alongside the app in Docker Compose, and the whole thing is built spec-first with GitHub Spec Kit. Recent work adds shared draft settings gating, selectable board themes, a live draft broadcast ticker, and a switch to Sleeper for player data.

Currently building: completed-draft persistence has shipped — a static, exportable draft-results view, a board-image export, and auto-completion once the final slot is picked all landed this month. In review now, not yet merged: an authenticated commissioner front door with durable "My Boards" (replacing the developer-only demo index), a protected AWS production CD pipeline — the deployment work this page used to call deferred — a draft-night reliability harness plus browser end-to-end and frontend test coverage, a formal contract for archiving completed drafts to cold storage, and continued broadcast-UX polish across the live board, pre-draft controls, and terminal completion flows. Per Linear, live pick analytics and market-consensus grades for the broadcast ticker remain queued up next.

Outcome

The repo is private while it's early, but the pace of work makes the case: completed-draft persistence shipped this month, and nine more pull requests are in review right now — a real commissioner front door, a production AWS deployment pipeline, and draft-night reliability testing across projector, seat, and browser sessions among them. This page will link out once there's something safe to share publicly.