ODYSSEE

Welcome to ODYSSEE, the automatic team distribution for "Praxis der Softwareentwicklung" at the KIT.

What's ODYSSEE?

ODYSSEE is the central administration tool for Praxis der Softwareentwicklung (PSE). It stands for:

Odyssee
Distributes
Your
Students (to)
Software
Engineering
Exercises

If you want to participate in PSE this semester, you need to register here. Afterwards, you can view and rate available projects. You can also form a group with fellow students if you want to work on the same project together.
Then, ODYSSEE assigns each student a project considering your wishes and other external constraints on a best effort basis. When the distribution is complete, you will be notified via email about your assigned project and team. You can also find this information when you log back in into ODYSSEE.

The kick-off event (Auftaktveranstaltung) gives further information about the projects. Also, don't forget to register for PSE, i.e. the TSE and the PSE exam, in the Campus system.

Current PSE PSE Winter 2022/23

Sign up start Friday, October 28, 2022 3:00 PM
Sign up deadline Monday, October 31, 2022 11:59 PM
Link to the official PSE homepage. Mail the organizers.

Projects (not final)

Hint: Click on a project title to view more information about the project.

At TECO, we're developing and end-to-end machine learning testing platform - Validaitor. Validaitor stores trained machine learning models and datasets so that people can write tests on these models. Validaitor currently supports performance, fairness and data drift tests as well as some monitoring capabilities. Validaitor is written in Python using Django as the web framework. As database, we're using Mysql and for asynchronous communication we're using RabbitMQ and Celery. You can learn more on www.validaitor.com


During PSE, you'll develop a real-time and interactive machine learning analysis dashboard that will be integrated to Validaitor. You'll develop the dashboard infrastructure as well as the functionalities like model explanation using SHAP and LIME, what-if analysis etc. When developing your solution, you'll be familiar with the existing codebase of Validaitor and extend it accordingly so that your dashboard works perfectly in coordination with the system. At the end of the project, you can expect to have a good understanding of Python, Django, webservices, machine learning explainability etc.

At TECO, we're developing and end-to-end machine learning testing platform - Validaitor. During your PSE, you'll develop its SDK using Python. Using this SDK, users will interact with the Validaitor server from inside their Python code. When developing the SDK, you'll also learn the current codebase of Validaitor and extend it by writing Rest APIs to communicate with your SDK. Validaitor is written in Python using Django as the web framework. As database, we're using Mysql and for asynchronous communication we're using RabbitMQ and Celery. You can learn more on www.validaitor.com


Validaitor stores trained machine learning models and datasets so that you can run tests on these models. Currently Validaitor supports performance, fairness and data drift tests. Besides, it also offers some monitoring capabilities. The SDK you'll write will support all of these capabilities of Validaitor and bring easiness of use for people who'd like to communicate with Validaitor inside their Python code. At the end of the project, you can expect to have a good understanding of Python coding and its ecosystem, how to develop SDKs, how to write webservices etc.