Employee Tech Documentation — an in-house replacement for spreadsheet IT

A custom Flask + React asset tracker with MSAL/LDAP auth, audit trails, and M365/UniFi integrations. Deleted the IT spreadsheets.

full company

Employees in system

M365 · UniFi · Zoom

Integrations

MSAL + LDAP

Auth surfaces

every mutation

Audit coverage

The IT function ran on a set of shared Excel files — who has what laptop, which phone, which software license, what access. Those spreadsheets were the reason audits took weeks and offboardings were incomplete.

I built a proper asset-tracking / employee-technology system in Flask + React. Authentication via MSAL for M365 users and LDAP for local AD. Every record is audit-logged. Integrations pull asset state directly from M365, UniFi, and Zoom so the database doesn't lie.

Stack

Python / Flask backend
React + TypeScript (Vite) frontend
SQLite with in-app schema-upgrade migrations
Session-based auth (MSAL, LDAP)
M365 Graph + UniFi + Zoom API connectors

Why custom

There are off-the-shelf IT asset management tools. Most of them are either too expensive for a site our size, too generic to integrate with the systems we actually use, or both. For a one-person IT shop, a lean custom app that knows exactly our integrations is a better investment.

What the system does

Tracks employees, their assets (laptops, phones, monitors, peripherals), and the license footprint per user.
Reconciles the tracked state against live API data from M365, UniFi, and Zoom — flags drift (license assigned in M365 but no user record; user record but no UniFi access; etc.).
Runs the onboarding + offboarding playbooks — it knows what to provision and what to revoke, with checks against each integration that the action completed.
Full audit log on every write — who changed what, when, from where.

Lessons

Building your own IT tooling is worth it exactly when the off-the-shelf option would require more customization than the custom build. The integration surfaces were the hard part; the CRUD was trivial. Picking the right data model — the relationships between employee, asset, license, access — took a couple of weeks of iteration and is why the system earns its keep.