INTERN - Support Engineer

Schneider Electric Systems
Industrial Automation - Global Customer Support

• Language: English / Dutch. All documentation is in English.
• 32 to 40 hours a week, 5 months.
• Location: Baarn Netherlands.
• Working from home is an option but the preference is to be in the office 75% of the time. Physical setups in the test lab can only be done in the office.

General overview

You will be working in an customer support environment in an industrial automation setting. Customers from the Europe/Middle East/Africa region report problems with their installed systems. Your job is to break down the customer complaint into manageable parts, work with the local engineers at site to get an overview of the installed software/hardware and then root-cause the reported problem.

For issues that cannot be solved at the customer site a test setup with similar hardware/software has to be made in the test labs in Baarn NL. Then various simulations need to be run to recreate the problem seen at the customer site. Once the problem is recreated a test protocol must be written and handed over to the development group. Once the development group has created the software fix it needs to be tested in the test lab to verify that the problem is indeed fixed, before handing it over to the end user.

Technical details

Customer installations are Distributed Control Systems (DCS), consisting of 10+ workstations, 10+ switches, and 10+ PLC type industrial controllers. Systems are located at large chemical or other industrial installations. Workstations are either physical Windows stations or Microsoft HyperV virtual machines. Network switches are setup in a specialised redundant setup using Rapid Spanning Tree protocol.

Recommended Technical Skillsets

• General Microsoft Windows operating system knowledge is a must have.
  Knowledge of Microsoft Active Directory, group policy setup, etc, is good to have.
• Knowledge of Microsoft SysInternals tools will help in diagnosing of problems. If not familiar with SysInternals then you will need to learn the basics using the computers in the test lab.
• Knowledge of Wireshark network analyzer will help in diagnosing of problems. If not familiar with Wireshark then you will need to learn the basics using the computers in the test lab.
• Knowledge of working with an oscilloscope and multimeter to do measurements on incoming data from the field is a nice-to-have.
• Knowledge of efficiently dealing with large volumes of logging data is preferred. At the moment Elasticsearch is used to process bulk data into dashboards and to do analytics.

People Skills

• Being able to communicate clear, effective, and professional is a skill that needs to be grown in this type of work.

Possible avenues of investigation

Apply LLMs or local AI to solves real world problems. Due to the sensitive nature of the customer data all AI models and Large Language models (LLM) must be run locally or on Schneider Electric approved cloud storage. At no point in time data is allowed to be uploaded to public AI providers. Can we use on-device models to help out with any of the following pain points?

1. Bulk data log file analysis

Customer systems have anywhere from 10 to 250+ devices. Each device produces logs files with info and error message 24/7.

At the moment Elasticsearch with Kibana dashboards is the tool used to investigate. Can other local tools bring measurable improvements?

2. Microsoft SQL server optimization across different workloads

The algorithms to run the industrial application are stored in complex SQL tables. We are always looking for performance improvements. What can be done to make the Microsoft SQL database performs better within the constraints of the hardware/software?

Added difficulty: each customer site runs it's own SQL database with it own loading parameters. So instead of one big database there are 100+ smaller, individual customer databases to be compared and fine-tuned. With SQL data from multiple customers, can a general optimisation scheme be made?

3. Network storm and congestion prevention

In these industrial applications available network bandwidth is crucial. Any network storms on the switch network need to be identified, root-caused, and changes will have to be made to prevent them from happening again.

Wireshark is the preferred tool to diagnose these issues. Elasticsearch with Kibana dashboards can also be used.

Added difficulty: with network data from multiple customers, can a general optimisation scheme be made?