The story

I work security at a data center, and our facilities worker is the person we go to when something physical goes wrong: broken lights, gates that need maintenance, grass that needs cutting. We talk a lot, and I kept hearing about the same problem. Vendors were showing up late or forgetting their visit completely.

He kept the visit schedule in an Excel sheet, and reminders only happened if he remembered to send them. My first idea was automatic emails through Microsoft. That was out of the question, because IT had the access locked. Online services weren’t an option either, since they couldn’t be trusted with company data.

So I needed something local. I came up with a Python app that runs on his machine and works from the spreadsheet he was already used to.

What I built

Vendor Manager is a small Python desktop app that sends the reminders by itself. Every day it:

1. IReads the visit schedule and finds tomorrow’s visits
2. IILooks up each vendor’s email in a second sheet, the vendor directory
3. IIISends a personal reminder from his own Gmail or Outlook account
4. IVWrites every email it sends to a log file

He didn’t have to change anything about how he works. He keeps updating the same spreadsheets and the app takes care of the rest.

The first time, a setup wizard points the app at his two files and connects his email. Gmail uses an app password, and Outlook goes through Microsoft Graph.

The control panel lets him pause it, resume it, or send the reminders right now, and he can read the log to see who got an email.

Real spreadsheets are messy, so I made the reader forgiving. It finds columns by their header name instead of their position, matches vendor names no matter the capitalization, and understands dates typed in different formats or saved as Excel date numbers. The email itself is a plain text template with placeholders, so he can change the wording without touching any code.

To hand it over, I packaged it with PyInstaller into three programs that run on their own: the setup wizard, the control panel, and a background runner that the scheduler starts every day. He doesn’t need Python installed. I also wrote unit tests for the Excel reader, the email sender and the settings, and they run on GitHub Actions every time I push.

What I learned

It felt great helping someone with the knowledge I have.

• ❧Limits shape the solution. With the cloud and outside services ruled out, a small local app was the most trusted option, not a second choice.
• ❧Build around how people already work. Keeping his spreadsheet as the source of truth is the reason he actually used it.
• ❧Handing it over is half the work. The setup wizard, the standalone programs and a log he can read mattered as much as the email code.
• ❧Small tools open doors. He was very happy with it and told everyone about the app, which helped the projects that came after.

The story

Badge access in the data center is requested through ServiceNow, and it works in a hierarchy. When someone gets a space, like a data hall, the request comes with a set of access codes inside it: the front gate, the front door, and the data hall itself. But in the system, the whole thing is just named data hall.

When door names change with different customers, the access gets mixed up, and sometimes time is the only thing that fixes it. I ran into some of these problems and reported them. Then I realized there were very similar problems that had become the norm, and people were using workarounds just to get their access.

I thought the problem was deeper. Property management was curious about what I could find, and they let me check every access in the building. I was overwhelmed. I needed something where I could track what to delete, what to correct, and the unique problem with each access code I found, so I could report them all at once.

That’s what this app was for. I wish I could share more about the problems I found, but it’s work information. What I can say is that this project meant a lot to me.

What I built

Access Manager is a simple web app with one job: keep the audit organized. Every access point is in one list, grouped by section, so I could go through the building step by step instead of working from scattered notes.

Every entry has a type, like data hall, MEP room, cage, rack, loading dock or office, and filters let me narrow the list by name, section or type right away.

• ❧Notes: every entry has its own note for the specific problem I found at that door.
• ❧Completed: once I check an entry, I mark it done and it moves out of the way.
• ❧Change log: every action gets a timestamp and is saved, so I always know what changed and when.

I built it with plain HTML, CSS and JavaScript, no framework. Supabase (PostgreSQL) stores the entries and the log and handles the sign in, so only signed in users can get in. It’s hosted on Netlify. Keeping it this small meant I could build it fast and start the audit right away.

What I learned

With the app tracking my progress, I went through over 130 access codes by hand. Other security officers saw that job as hell. For me it gave me so much joy, and property management was impressed that I made a tool for the problem instead of just living with it.

• ❧A small tool can carry a big job. Notes, a done list and a log were all I needed to turn an overwhelming audit into steady progress.
• ❧Normal isn’t the same as fine. The workarounds everyone was using were a sign of a deeper problem.
• ❧Report it all at once. Having every issue in one place let me hand over the full picture instead of one complaint at a time.

Tackling problems with my own tools made me feel like a detective, and that opened my imagination to what I can do as a software engineer.

The story

After going through the access code problems, the access problems in my building dropped a lot, and I gained a lot of knowledge about data center access. The property manager started asking me about all types of access and where they are. In one conversation she asked if there was an access map that showed all the access codes.

There wasn’t one. That matters, because some access names still had problems, and others weren’t clear about where they actually give access. On the property side, knowing that is essential before approving anyone.

Property gave me a floor plan to write the names on for their reference. But I felt it would get cluttered, especially since one access can open doors in different places in the building. So I thought of making a map app instead. First, you search for the access name and it takes you to it on the map. If that access is in different places, a dynamic arrow takes you to each one, with a count. I also needed to be able to add, remove and edit points for future changes.

I knew the map would expose another set of problems at the same time: how access is categorized, and whether it makes sense to both the person requesting it and the person approving it, in a way that stays maintained.

Then for a month, me and my map went to every single door in the building to see how the spaces were cut. I asked every team, from engineers and technicians to janitorial staff and customers, about these rooms. I was able to finish my whole building.

What I built

Access Maps turns any floor plan into an interactive map. You load the plan as a PDF and pin access points, which I call nodes, anywhere on it. You can move around and zoom in like a normal map app.

The main feature is search. You type an access name and the map jumps to it. Nodes with the same name are grouped and counted, because one access code often opens doors in a few places. When there’s more than one, a dynamic arrow points you to each place in turn.

Nodes can be added, moved, edited and removed at any time, so the map keeps up when doors and customers change.

Under the hood it’s a React app built with Vite, and PDF.js renders the floor plans. I split the code into small modules that each do one job: the viewport (moving and zooming), the map layer, the node layer, the search engine, and a data layer that talks to Supabase. Supabase also handles sign in, and if you leave it idle, it warns you and then signs you out. It’s hosted on Netlify.

What I learned

The map became a reference for property, a training tool for security, and a great place to look when we face access problems.

• ❧Software is only as good as its data. Building the app was only half of it. Making it accurate took a month of walking every door and talking to every team.
• ❧The people who use a space know it best. Engineers, technicians, janitorial staff and customers all knew things about rooms I couldn’t figure out alone.
• ❧A good tool shows you the next problem. Mapping access brought up bigger questions about how it’s categorized and whether requesters and approvers read it the same way.
• ❧Plan for change. Doors and customers change, so editing the map had to be as easy as reading it.