Projects We involved
Real time Scheduling Project
In this project, a real-time job scheduler for 5G-RAN in a Base-Band Unit (BBU) was developed in the form of a PoC (Proof of Concept) for a hierarchical inter-cluster and intra-cluster scheduling to not only efficiently schedules all tasks but also considers load balancing, dependencies, timing and other system requirements. We started with identifying the parameters, formulating the problem, then proposing state-of-the-art algorithms to deal with the optimization problem, and finally analyzing the result.
Routing Optimization Project
In this project, we developed an optimization software service for solving a Constrained Vehicle Routing Problem with Time Windows (CVRPTW) based on VROOM in Python and C++. Multiple ML and Neural Networks models are also applied to predict the key parameters such as service time and vehicle capacities. CVRPTW is a keystone problem in (multi) hub-based last-mile deliveries to dispatch tons of items with the minimum cost in a timely fashion. A simple demo of the service is available in DockerHub.
Food delivery Project
In this project, an AI-based optimization software service to solve a Travelling Salesman Problem with Time Windows (TSPTW) was developed based on Google OR-Tools in Python. We also developed multiple ML-based features to predict the key parameters, including preparation time and service time. The goal of our optimization service is (1) to maximize end-users satisfaction by delivering food in the expected time and (2) at a minimum cost per delivery while respecting the constraints. A simple demo of the service is available in DockerHub.
