# HumaPace · DTSC Collection · DTSC

> Match needs in the field with skills and availability, then compare several responses to an absence. The decision remains human.

Source: https://www.dtsc.be/en/brain/projects/humapace/

DTSC

BRAIN / Projects & products

DTSC Collection

DTSC Collection, volume 15

# HumaPace

Anticipate. Organise. Adapt.

Scheduling teamsHospitalEducationSMEs

DTSC Collection

HumaPace

Workforce planning

S. Aulotte

No. 15

15

HumaPace

DTSC

LLM required

49 + 37

tests

sectors

CP-SAT

Leaflet

Human decision

HumaPace proposes assignments, explains the changes and lets you compare three objectives, stability, mobility and balance, before touching the official schedule. An independent validator checks every proposal. Needs that cannot be covered remain visible: the computation invents neither skills nor capacity.

sectors, one shared engine

49 + 37

backend and frontend tests

42

views checked at 1,440, 390 and 320 px

LLM required to schedule

## Technical cross-section

Layers: interface, business, computation, data, base.

Interface

Next.js · React · TypeScript

management, team leader, employee: three experiences

Accessibility

keyboard, accessible dialogs, reduced motion

Business logic

FastAPI API

scopes enforced on the API side, not by hidden buttons

Versioned proposals

an outdated proposal can no longer be applied

Computation

OR-Tools CP-SAT

hard constraints, coverage, weighted objectives

Independent validator

every assignment rechecked outside the solver

Data

PostgreSQL · SQLAlchemy

versioned snapshots, Alembic migrations

Audit

actor, action, version and date; neither password nor token

Base

Docker Compose

ports bound to the local loopback, database with no published port

GitHub Actions CI

pytest, typing, containers, Playwright

Backup

restore tested on a separate database

## Preview

Control centre: coverage, capacities, reorganisation to prepare. Fictitious staff.

## On the hospital side

Hospital laboratory: eight departments, seven-day rotations, coverage department by department. Fictitious staffing.

## What this volume demonstrates

Optimising under constraints

Qualifications, availability, ceilings, locked commitments.

Separating computation and decision

The engine proposes, a human validates, the history keeps the record.

Delivering continuously

A CI that builds the containers and replays the journeys.

## Under the binding

Next.js

React

TypeScript

Leaflet

FastAPI

OR-Tools CP-SAT

SQLAlchemy

Alembic

PostgreSQL 17

Docker Compose

GitHub Actions

Playwright

## Acknowledged limits

Local demonstrator with fictitious staff data. SSO, isolation between clients and operating guarantees belong to the target V1.

All projects and volumes
