Cases 2020
At the current stage we opened a call for industrial challenges from companies. Here are several examples of challenges:
You are welcome to submit a challenge at a.sadovykh@innopolis.ru
e-Citiz - feasibility of migration: Jenkins to GitLab CI
Context: e-Citiz is a product by a French company Softeam (DocaPoste brand). e-Citiz is a process automation solution applied in many government organizations in France and abroad. e-Citiz provides digital single-window services such as requests of id documents, school enrolment, voting and etc.
Goal: The e-Citiz platform and services are developed in Java in a continuous integration cycle. The previous versions of the platform we integrated with the use of SVN and Jenkins, we are progressively moving to Git (Hosted by Gitlab) and GitLab CI. For the analysis, we would like to evaluate the feasibility and cost to move our Jenkins infrastructure to GitLab CI.
For the experiment we will intend to provide:
List of plug-ins currently in use with Jenkins
We expect:
A report on feasibility and cost - supported by an experiment.
The analysis problem may be refined during interviews.
More info: List of Jenkins plug-ins.
Loudplay - metrics for quality assessment
Context: Loudplay is a cloud gaming business. The service allows playing in the resource-demanding video games in the cloud. Our company is one of the few Russian companies with a commercial demand at the world level. Moreover, our solution is recognized by the leading tech giants such as Huawei, Nvidia, and ASUS.
Goal: Our products and services scale rapidly that makes us adjust our development and quality assurance practices. For the CASE in Tools hackathon, we propose to analyze metrics to evaluate the quality of our gaming service. In particular, we are interested in screen capture quality and playability of cloud gaming.
As the starting point, we recommend researching MOS (Mean opinion score) of video and audio quality and apply this quality model to the cloud gaming industry. This model could be used as an estimation tool for telecommunication companies to evaluate the profit of partnership with us. Ideally, we expect a research document with metrics and formula for calculating the GMOS (Gaming mean opinion score). An additional plus will be if students will provide a way of capturing these metrics.
For the experiment we will intend to provide:
Server by Loudplay
Client app of the current version
Research documents about MOS provided by Huawei
The analysis problem may be refined during interviews.
MIPT - Continuous Documentation Delivery
Moscow Institute of Physics and Technology is a leading university in cutting edge digital technologies and a competency centre in artificial Intelligence for national technology initiative (NTI).
Context: MLDev project aims at developing open source tools to facilitate repeatable data science experiments, presentation and visualization of experiment results for students, data scientists and researchers.
The MLDev software provides the following features to help with automating machine learning experiments:
Run repeatable experiments in Jupyter notebooks in Google Colab or PaperSpace
Write your logs and results directly to your Google Drive
Receive notifications via Telegram about the exceptions while training your model
Track experiment logs and result via TensorBoard on Google Colab
A web service to access your trained model
MLDev also provides custom experiment templates and extensions that can be used to automate tedious work in data science university courses.
Goal: MIPT investigates feasibility of the tool that would harvest and assemble documentation for a given project from the project repository and publish it on the web automatically as part of the Continuous Delivery process.
The workflow and solution to be evaluated as depicted in:
The problem for the analysis is:
Analyse the feasibility of the workflow
Evaluate the efforts required for setup and regular updates
The expected result is:
Presentation of the analysis results based on an experiment
Discussion of implementation alternatives
Proof-of-Concept that demonstrates the feasibility of selected approach
The analysis problem may be further refined during the interviews.
RedFabriQ - Analysis of low-code platforms for Web Sites generation
RedFabriQ is a French startup-studio with a software & hardware culture. Among the projects that RedFabriQ are promoting are related to what are referred to as low-code platforms.
Context: Low-code development platform is an application that provides the Graphical User Interface for programming and thereby develops the code at a very fast rate & reduces the traditional programming efforts.
The team will experiment with several low code platforms by implementing a basic toy Web site for a Digital Library. Based on this analysis the team will deliver an analysis presentation.
Tasks: Try to implement a CRUD application to manage a library (authors, book reference details, title page) on 6 low-code platforms and try to find meaningful criteria to compare them.
Example low-code platforms you can pick: Adalo, mendix, bubble, retool, glideapps, honeycode
You’re encouraged to choose ~2 additional low-code platforms.
The analysis problem may be further refined during the interviews.