On a mission to enable building ethical, explainable, responsible and sustainable AI
About BUILDATHON
Buildathon 2024 powered by PreProd Corp is an exclusive 2 days (1st & 2nd August 2024) live hackathon event designed for VIT Bhopal students.
The outcome of this event will be an AutoML user interface capable of automatically training, testing, validating, and delivering ready-to-use supervised AIML models for any dataset selected by the user. As a participant, you will learn to design, develop, deploy, tune, and maintain complex AIML pipelines within a limited timeframe. You'll also gain experience working in a live agile team, simulating real-world technology roles such as frontend developer, backend developer, data scientist, Python developer, and more.
Happy building !
PROBLEM STATEMENT
In a hypothetical scenario, PreProd Corp requires to design and deploy a freely available UI that can train AIML models using any dataset that qualifies certain conditions i.e. build an AutoML.
As a BUILDATHON participating team you are expected to build a frontend & backend that could allow the end user to input a dataset as csv or excel file. Perform basic cleansing on the data i.e. ensure the dataset provided should be compatible to be used to AIML model training. Use the data to train at least on 5 different popular AIML models. The AutoML tool should be able to provide the pickled file of the all or best performing trained AIML model to the user.
Overall product
Build a frontend - backend AutoML solution that can ingest any dataset ( supervised learning supported, having a maximum of 5000 rows and 10 columns ), designed AutoML should train AIML models in backend and output the pickled file for the trained models. The design should look modern and easy to use - navigate through. Please refer the AutoML wireframes
Input data
Dataset in CSV or XLSX or XLS format, should have predictor & predicted columns i.e. a supervised learning dataset, having a maximum of 5000 rows and 10 columns.
AIML models
AutoML should have any 5 or more of the following AIML algorithms:
SVM
Decision Tree
Bagging
Random Forest
ADA Boost
XG Boost
Stacking
Fully connected neural networks
AutoML should display the below performance metrics for each trained algorithm:
Accuracy
RMSE
AUTOML WIREFRAMES
You are expected to develop a clickable app using any technology as a front end for python based AIML model training on the backend. Please refer to the below wireframes as a reference to build your product which looks and performs similar to the below designed wireframes.
REGISTER YOUR TEAM FOR BUILDATHON
Please read the below rules for forming and registering a team:
One team should have a maximum of 5 members
All team members should be current students of VIT Bhopal and all members should be present in the campus for both the days for this event
Team should be self sufficient to develop this app
Each team should come up with a quirky name
One member per team to scan the QR code to complete the registration
Do not register for this event if you are not currently pursuing your degree at VIT Bhopal
OR
24 Jul
Web Launch
30 Jul
Registration Deadline
1 AUG
Buildathon Start
2 AUG
Buildathon End
REWARDS & RECOGNITION
Top 3 winning teams will get an internship opportunity (no stipend) with PreProd Corp for the duration of 2 months where you will get experience to build data engineering, XAI and MLOps pipelines. All participants will receive a certificate of appreciation.
2 teams on special consideration from the hackathon evaluators, who were very close to complete the challenge but could not complete the challenge will be given an opportunity to obtain mentorship from our technology team to complete. All participants will receive a certificate of participation..
All the participants will be invited for 4 virtual sessions where our technology team will mentor and teach them to build an AutoML from scratch. All participants will receive a certificate of participation.
Winning teams
PROMISING TEAMS
