YourTutor

Intro
Learners description
Learning progress
Model findings & learning way
Learning summary

Inspiration

Multiple news about struggles of education systems with transition to online learning, especially for students who prepare for exams or those who don't have access to stable internet connection. We wanted to use our knowledge in cutting edge technologies to help resolve multiple issues with current approaches to remote education.

What is does

Our solution proposes active online learning approach contrary to currently present passive learning, shifting the center of interest from content to learner. We developed a framework for training personalised models for each student, which would guide them in their education by assigning tasks of appropriate difficulty level and perform tests to assess their improvement in different areas. Such system could improve efficiency of both students and teachers by shifting their focus to areas in need. Moreover, our solution can be used for various different use cases outside of public schools, such as private language schools or private tutors.

How we built it

We used a Reinforcement Learning algorithm, developed it in Python and filled with custom mechanisms, regarding the optimisation direction and modelling of the students by defining their properties and randomly sampling them to obtain a realistic simulation. We performed the simulation on two sample students, which could be one of three defined types of learners and would study three different subjects.

Challenges we ran into

Customisations of the learning path so that the algorithm goes in the right direction.
Choosing appropriate revenue streams and coming up with business model that would assume profit.
End product design turned out to be difficult for a customisable platform.

Accomplishments that we're proud of

We are proud of what our final simulation page looks like, especially since majority of our team are machine learning engineers. We delivered estimation of deployment and maintenance cost for public customers and did initial market research. Our machine learning model learned to recognise which type of content (visual, auditory or textual) resulted in largest skill gain for each student. We are very happy with our team work and our cooperation with mentors, for which we are truly grateful.

What we learned

We learned that focusing on the product itself is not enough - we need to be able to pass the knowledge and present it in a comprehensible way. We learned the value and mechanisms of market research and finding revenue streams, which we never did before. We also spent some time to educate ourselves on pitch preparation with a strict time limitation. We also gained a lot of knowledge by experimenting with our teacher-learner environment and are full of conclusions for future development.

What's next for YourTutor

Improvement of our AI model, research for education content providers, platform requirements specification with help of an expert in the field of education, establishing relationships with potential clients, working on design and comprehensible concept transfer approach.