bruno.exePROJECTScontact.exe
ATLAS

</atlas>

ATLAS

Internal systems and tooling a logistics company still runs its operations on.

8 weeks

built solo, wireframes → production handover

live

data feeds in place of a once-a-day spreadsheet update

auto

weekly report. Replaced a 2+ hour manual report every Monday

SPECIFICATIONS

ROLESOLO BUILD
DATE2025
TYPEINTERNAL SYSTEMS · AUTOMATION
STATUShanded off
STACKinternal tools · automation · typescript · real-time · role-based access · automated reporting
LINKSnone public
CLIENTLogistics client
SOURCEhanded off to client, NDA

A logistics company was running its operations on spreadsheets updated once a day.

== WHAT IS THIS ==

The internal systems and tooling a logistics company runs its operations on, still in use today. Built solo in eight weeks: live data feeds in place of spreadsheets updated once a day, role-based access for three roles over one dataset, and a weekly report that is assembled and sent on a schedule instead of by hand every Monday. Handed off to the client under NDA.

== </stack and architecture> ==

internal tools · automation · typescript · real-time · role-based access · automated reporting

Everything runs on one dataset. Live data feeds push changes into it as they happen, so nobody updates a sheet or waits for the daily refresh. Access is role-based and enforced at the data layer: three roles each get their own view and permissions over the same data rather than three copies. Reporting is a scheduled job: once a week it compiles the last seven days of metrics into a report and emails it to management.

== </how it works> ==

Live data feeds

Changes are pushed into the systems as they happen, in place of spreadsheets that were updated once a day.

Automated weekly report

A scheduled job compiles the week’s metrics into a report and emails it to management, replacing a report that was put together by hand every Monday.

Role-based access

Three roles each get their own view and permissions over the same underlying data, enforced at the data layer.

Handed over, still in use

Code, designs and documentation went to the client’s internal team at the end of the eight weeks. The team still runs its operations on it today.

== </key decisions> ==

DECISION 01

The weekly report became a scheduled job first. It replaced a job someone did by hand every Monday. It was the least glamorous feature and the clearest time saving.

DECISION 02

Live updates instead of polling. Only changes are pushed, which keeps everything current and the load on the database low.

DECISION 03

One dataset, three roles. Permissions are enforced where the data lives, so a new view does not mean a new copy of the numbers to keep in step.

== </the problem> ==

A logistics company was running its operations on spreadsheets updated once a day. The team could not see what was happening during peak hours, the weekly report for management was put together by hand every Monday, and three roles needed different access to the same data.

== </the approach> ==

Build the internal systems the operation runs on, and automate the manual steps rather than add another screen. I designed and built them solo in eight weeks, from wireframes to production and handover: live data feeds in place of the spreadsheets, role-based access for three roles over one dataset, and a weekly report assembled and sent on a schedule. The work was handed to the client's team under NDA, and the team still runs on it today.

role & context

Solo freelance engagement for a logistics company. I designed and built the internal systems and tooling its operations run on, delivered them in eight weeks and handed them to the client’s internal team, which still uses them today. Under NDA, so the client, the data and the exact stack are not described.

Codebase handed off to the client under NDA: architecture, process, and outcomes are described with client details generalized. No repository or internal screenshots are public.

== </the story> ==

Atlas is the set of internal systems and tooling a logistics company runs its operations on, and the team still uses it today. Before it, operations ran on spreadsheets updated once a day. Knowing where things stood meant waiting for the next update, and the weekly report for management was put together by hand every Monday.

The brief was to take that manual work out of the week: feed the operation live data, give each of three roles its own access to the same numbers, and produce the weekly report without anyone assembling it.

I designed and built it solo in eight weeks, from wireframes to production and handover. The code, designs and documentation went to the client's team under NDA, so the details here are kept general.

== </what was learned> ==

>

The automatic report saved the most hours. The least glamorous feature often matters most.

>

Pushing live changes beats asking for them over and over: faster for people, lighter on the system.

>

Automating a task starts with finding who does it by hand today and when. The Monday report was the obvious one once someone said it out loud.