We started our journey by understanding the importance of a job recommendation system based on the skill set of the user. A system which can, not only recommend the job but also highlight the necessary skill set needed for the recommended job, which helps the user to learn more on the skill set
What is the solution
A Flask web application backed by powerful deep learning model and ESCO-API to recommend job based on user skill set along with recommending necessary skill set needed for the recommended job
How we build it
- We first used the "occupation_en.csv" data provided by the hackathon manager but failed to build model due to overfitting caued by small dataset.
- We found a bigger dataset from kaggle and used that to train our model on Google Colab which has given 63% validation accuracy
- Then we used that model to build a Flask web appilication so that user can interact with it on Google Colab.
- We integrated the ESCO-API in to the web application. Which means after the job recommendation, the ncessary skills also will be returned for that specific job.
Running the Flask web application
- Install anaconda python on your system
git clone https://github.com/ADARSHULTIMATE/ideahack.git
conda create -n recommendation pip python=3.7
conda activate recommendation
pip install -r requirements.txt
Training the model
The challenges we faced
- Finding best hyper parameter to train the model.
- Creating a dictionary to map the predicted job to ESCO-API URI.
- Building the web application and loading the model inside it
Future of our solution
- Building more powerful AI model and better prediction accuracy
- Creating a database for easy identification of skill set for recommended job.