Graduation assignments Production and Logistics

2018-01 Two MSc assignments (IEM and CS) for IBM on Machine Learning for after-sales service supply chains

Assignment 1: Focus on machine learning (student Computer Science with basic knowledge of Supply Chain Management)

Assignment 2: Focus on operational planning of after-sales service supply chains (student IEM-PLM with basic knowledge of Machine Learning) 

Remarks:

  • Start of assignment: As soon as qualified candidates are available.
  • This project is intended as multidisciplinary “duo-assignment” of a student IEM and a student CS, so both students should be available in about the same time frame.
  • This project is a follow-up of a first exploratory project, where an IEM student and a CS student built a data infrastructure and performed some first analyses. You can use the infrastructure available at IBM for this follow-up project.  

The Company

IBM is a company of ± 400,000 individuals who do, in new ways, what IBM-ers have done for a century: invent technology and apply it to business and society on a global scale to make the world work better. Today, we create and integrate hardware, software and services into cognitive solutions that enable enterprises, institutions and forward-thinkers around the world to succeed. IBM operates in more than 170 countries and enjoys an increasingly broad-based distribution of revenue, grouping markets by common growth characteristics, not location.  

Within IBM, the Service Parts Operations (SPO) organization is responsible to provide the service parts required for IBM hardware maintenance on IBM logo and non-IBM machines. SPO has a dense worldwide logistics network that can supply spare parts to customer site with a service level ranging from second business day to within 2 hours. IBM has located their Europe, Middle East and Africa headquarters for SPO in the Netherlands back in the late 80’s. Since that time major innovations and transformations took place relative to IBM’s SPO, which has resulted in major improvement in terms of reduction of spare parts inventory, improvement of service levels and reduction of logistics cost. In the early 90`s, IBM parts operations in Europe, Middle East and Africa was organized at a country level, with country specific processes and inventory ownership at country level. This mode of operation was changed during the last decade of the 20th century into an operation managed centrally out of the Netherlands for the whole of Europe, Middle East and Africa. This was supported by introducing a run-once do once single instance IT system called CPPS.  

Since the beginning of 21st century SPO has been on a path of Globalization of processes and the related organizational setup. For the key processes in IBM’s Service Parts Operations this has been implemented. Besides the introduction of worldwide common processes and the associated simplification, IBM has invested substantially in improving and enhancing the planning of the spare parts in all of its life cycles. Key achievements here are the introduction of the Neighborhood planning system combined with an enhanced spare parts allocation algorithm that supported the concept of High Availability Time based services. This solution links customer entitlement (contract information) with logistics execution. Significant investment has been put in the optimization in other processes as well, like asset recovery, reutilization etc. In 2003 IBM has been awarded the annual price from the VLM relative to IBM’s advanced way of dealing with reuse of parts and reverse logistics capabilities in Europe.  

IBM SPO's ambition is to create a highly reliable and flexible service parts management service that delivers the requested level of service with optimized inventory and cost levels. 

The assignment 

This assignment is related to the operational control of the service parts supply chain. Parts planners are making tactical decisions based on the actual status of the supply chain (actual stock, usage, demand patterns, network requirements). It is important that upcoming issues like understocking or overstocking are identified early, such that preventive actions can be taken. Other potential areas of proactive actions include (i) identification of delays in the physical distribution process having serious consequences, such that corrective actions can be invoked, (ii) reverse logistics, i.e. whether to return and repair failed parts that have been removed from the assets, and how to route them with which priority level. This is all part of the service control tower concept, aiming to co-ordinate operational control actions in a global service supply chain. To this end, IBM has various applications for data analytics that generate signals upon which actions may need to be taken, such as Servigistics Review reasons and  the Entercoms information portal. http://www.entercoms.com/ ). Such tools are used both for the IBM Logo supply chain, as well as for the operation of other service supply chains that are managed by IBM (e.g. Lenovo). 

Within the ProSeLoNext project, IBM Service Parts Operations has focused on research in the area of Machine Learning with as central question: "Can we develop a machine learning solution for automating parts planner's workload?". In a first Master Thesis, two MSc students have implemented the infrastructure required to enable Machine Learning and applied it to one specific area of planning decision.  

In a next step, IBM wants to apply this infrastructure to different areas of planning decisions and find out in which areas the Machine Learning leads to behaviour that is sufficiently predictable to allow the planner's activity to be fully automated. This will require:

  • understanding of the various planner activities
  • selection of the relevant data in the decision
  • measuring the predictability of planner's behaviour
  • understanding why certain areas are more conducive to Machine Learning than others

For IBM the aim is to actually automate certain workload if possible, for the broader research project it is important to understand where Machine Learning could be applied. 

You will find good support of the people at IBM Amsterdam to help you get settled and on your way, but the assignment also requires the student to be self-managing. It is not necessary to work at the IBM premises daily. 

Contact persons:

If you are interested, please send a CV and motivation letter to: