Lantek – Intelligent Project Optimisation
Lantek
Sector: Industry
Business Case
The problem is the efficient distribution of tasks that arise in a project managed with the agile method. As the sprint progresses, employees are assigned tasks based on their availability and skills.
Objectives
Seek optimal scheduling in an automated manner to minimise total costs, considering task interdependencies and resource constraints. This allows optimising planning and resource allocation, reducing costs and increasing efficiency in project management.
Use case
A study of the most suitable ML and optimisation algorithms to address the problem of task sharing with scheduling was carried out. In this study, an optimisation algorithm with traditional conditions and a genetic algorithm with constraints was implemented.
Infrastructure
Cloud
Technology
Machine learning and deep learning
Data
Synthetic data generated with the PSPlib and Imopse library. They fit any organisational problem and are of standard format.
Resources
Personnel: 4 data scientists, 1 MLOps engineer. Infrastructure: Databricks workspace, 12-core processing cluster.
Difficulties and learning
Research and implement state-of-the-art optimisation processes with constraints for task scheduling on a calendar.
KPIs (business impact and metrics of the model)
Distributed 1500 tasks within 3 minutes while adhering to the schedule constraints. Can be scaled to process multiple projects in parallel. Executed in case of calendar or personnel changes.
Funding
Private with own resources. Dataton Euskadi 2023
Collaborators, Partners
Private with own resources. Datu(a). BAIC
