Attendance Data Consolidation Platform

Problem

Attendance information existed in several disconnected places — different sites kept their own logs, in different formats, updated on different schedules. Getting a single answer to "who was on site, where, and when" meant manually collecting and reconciling several separate sources, and the answer was already out of date by the time it was assembled.

Constraints

The existing per-site logging habits couldn't be ripped out overnight — people needed a low-friction way to keep logging the way they already did, while the data itself got consolidated centrally. Management needed a single live view, but site-level staff needed their own site's data to stay simple and familiar rather than being handed a complex new system.

Approach

I built a Go backend with a small set of well-defined ingestion endpoints that each site's existing logging habit could feed into with minimal change, and a central Postgres schema that normalised every source into one consistent attendance record shape. A responsive React frontend gave management a single live dashboard across every site, while preserving a simple site-level view for the people actually doing the logging.

Outcome

Management gained a single, live, cross-site view instead of a manually reconciled report that was always somewhat out of date. Site-level staff kept a workflow close to what they already knew, which meant adoption didn't depend on retraining everyone at once.

What I'd do differently

Normalising at ingestion time was the right call, but I'd invest earlier in a clear reconciliation/conflict story for the (rare but real) cases where two sources disagreed about the same person at the same time — early on this was resolved ad hoc rather than by an explicit, documented rule.

Architecture

The core idea: normalise once, at the point data enters the system, so every downstream consumer — the management dashboard, the per-site view, any future report — reads from one consistent shape instead of re-interpreting several different source formats every time.